0 INFO-VAX	Sat, 27 Jan 2001	Volume 2001 : Issue 53      Contents:F 64 bit systems with small memory, was: Higher sized memory for DS20[E]" A help message to Andrew Harrison.& Re: A help message to Andrew Harrison.& Re: A help message to Andrew Harrison. ACP-QIO and VBN  Backup Problem...  Re: Backup Problem... M Compaq annual meeting of analysts and investors - VMS to feature allegedly... 4 Re: Compaq: A simple, affordable clustering solution CTRL Z from a script Re: CTRL Z from a script Re: CTRL Z from a script Re: CTRL Z from a script Re: CTRL Z from a script Re: CTRL Z from a script Re: CTRL Z from a script Re: CTRL Z from a script Re: CTRL Z from a script Re: dec 3000 300 crashing  Re: dec 3000 300 crashing  Re: dec 3000 300 crashing  Re: dec 3000 300 crashing % Re: DEC TCPIP V5.0 problem with Linux   difference between EV6 and EV67?$ Re: difference between EV6 and EV67?$ Re: difference between EV6 and EV67?$ Re: difference between EV6 and EV67?$ Re: difference between EV6 and EV67?$ Re: difference between EV6 and EV67?$ Re: difference between EV6 and EV67? Examine log name on other node? # Re: Examine log name on other node? # Re: Examine log name on other node? # Re: Examine log name on other node? # Re: Examine log name on other node? # Re: Examine log name on other node? # Re: Examine log name on other node? # Re: Examine log name on other node? # Re: Examine log name on other node?  Re: fast_io_copy.c questions Re: fast_io_copy.c questions  Re: Filename wildcards under VMS  Re: Filename wildcards under VMS Filename wildcards under VMS FTP to GDG fails Re: FTP to GDG fails$ Re: ftp'ing savesets from W2K to VMS$ Re: ftp'ing savesets from W2K to VMS$ Re: ftp'ing savesets from W2K to VMS$ RE: ftp'ing savesets from W2K to VMS Future of VMS and Alpha = HELP, Is anyone using VMS as internet SMTP/POP server via UCX A Re: HELP, Is anyone using VMS as internet SMTP/POP server via UCX A Re: HELP, Is anyone using VMS as internet SMTP/POP server via UCX A Re: HELP, Is anyone using VMS as internet SMTP/POP server via UCX  Internet mail <> OpenVMS Mail ! Re: Internet mail <> OpenVMS Mail 6 Re: Is Compaq dropping support for vms on vax platform$ Re: It's the end for VMS get off now  It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: It's the end for VMS get off now$ Re: Linux worm and RedHat 7.0 broken$ Re: Linux worm and RedHat 7.0 broken$ Re: Linux worm and RedHat 7.0 broken, Re: Memo:  Re:  Memo:  Help! Zip/Unzip error. Re: Midrange I/O with VMS WAS: DS20 - Slow I/O Re: Moving Datatrieve files  Re: Moving Datatrieve files  Re: MPACK/MUNPACK images Re: MPACK/MUNPACK images Re: MPACK/MUNPACK images! Re: New OpenVMS Times now on line ! Re: New OpenVMS Times now on line ! RE: New OpenVMS Times now on line ! Re: New OpenVMS Times now on line " Re: Printing problems in TCPIP (2) Re: Problem with decnet copy Re: Problem with decnet copy Re: Problem with decnet copy1 Re: Question: Info. Needed On SHD51-ZZ Drive Box. 1 Re: Question: Info. Needed On SHD51-ZZ Drive Box. 1 Re: Question: Info. Needed On SHD51-ZZ Drive Box. ; Re: Rdb and EV6 (was: Re: difference between EV6 and EV67?) P Re: Rehash of Andrew's Java on OpenVMS FUD.  (was: Re: GS160 hardware     questi- Re: RMS bug, possible triggered by autoextend - Re: RMS bug, possible triggered by autoextend  Re: samba 2.0.6  Re: samba 2.0.6  Re: samba 2.0.6  Re: samba 2.0.6  Re: samba 2.0.6  select() and getservbyname()  Re: select() and getservbyname()  Re: select() and getservbyname() Re: SGI x Itanium x Cluster  Re: SGI x Itanium x Cluster  Re: SGI x Itanium x Cluster  Re: SGI x Itanium x Cluster  Re: SGI x Itanium x Cluster  Re: SGI x Itanium x Cluster  Re: SGI x Itanium x Cluster  Re: SGI x Itanium x Cluster 3 Re: Simple DECnet-IV Tasks Impossible under Phase-V 3 Re: Simple DECnet-IV Tasks Impossible under Phase-V 3 Re: Simple DECnet-IV Tasks Impossible under Phase-V 3 Re: Simple DECnet-IV Tasks Impossible under Phase-V 3 Re: Simple DECnet-IV Tasks Impossible under Phase-V ' Re: Source for used StorageWorks disks? ' Re: Source for used StorageWorks disks? ' Re: Source for used StorageWorks disks?  Re: Telnet, yes; HTTP, no ) Re: VMS 7.2-1 Bugcheck In DECW$REINIT.EXE ) Re: VMS 7.2-1 Bugcheck In DECW$REINIT.EXE 7 Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken) 7 Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken) 7 Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken) 7 Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken) 7 Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken) 7 Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken)  Re: VMS writable CD on Win2K?  Re: VMS writable CD on Win2K? 0 Volume Shadowing vs. RA7000 (WAS: DS20 Slow I/O)4 Re: Volume Shadowing vs. RA7000 (WAS: DS20 Slow I/O)4 Re: Volume Shadowing vs. RA7000 (WAS: DS20 Slow I/O)- Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open - Re: Window for Alpha success still seems open ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? ' Re: Writes flushed to Disk on Dismount? : Re: [Q]: Resetting privileged password on a DECserver 200?  F ----------------------------------------------------------------------   Date: 26 Jan 2001 17:18:04 GMT2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)O Subject: 64 bit systems with small memory, was: Higher sized memory for DS20[E] , Message-ID: <94sbgc$b7o@gap.cco.caltech.edu>  _ In article <200101260747.IAA19959@sinet1.fom.fgan.de>, Rudolf Wingert <win@fom.fgan.de> writes:  >Hello,  > H >at spring last year I asked for higher sized memory (2GB and 4GB optionJ >like ES40) for the DS20. Someone did answer, that at the end of year 2000L >we could see this one. Today I did have a look to Compaqs  Webpage. NothingK >to see about a higher sized memory. We work in the section of image under- I >standing. So we need workstations like the DS20E with high memory option H >(best would be >10GB memory). Does anybody know something about CompaqsI >plan with the memory of DS20 (1GB SDRAMs are state of the art)? Can hope ' >to hear something new early this year?   H I can't answer this question but it does bring up a problem that's been F bothering me for a very long time - since the Alpha appeared in fact. I Unless you do a tremendous amount of double precision floating point work J the only other reason to prefer a 64 bit cpu over a 32 bit one is to allowK direct addressing of large amounts of memory, that is, memory > 4Gb (32 bit I unsigned).  I can think of half a dozen instances off the top of my head  G where the data is largely static but must be randomly accessed at high  F rates.  (That is, load the data once and a while into memory, and READI access it at tremendous rates.)  For instance, in genomics work, a lot of J the search programs only work well when the databases are stored in memoryH because in many cases it's necessary to read through the entire databaseG once for each query.  A more commonly encountered application would be  E something like a phone directory service  (think www.superpages.com). G As in the genomics case the data is largely static, and here the search D times could be reduced from milliseconds (with disk accesses) to subG microsecond (memory access).  Also much of the perceived IO performance D that one obtains on Linux, NT, and pretty much every modern OS (VMS I excepted until XFC arrives) comes from file caching - the bigger the file J cache, the less likely the system will be to have to slow down and do realJ disk accesses.  (Let's leave the data integrity question out of it for now! - that's a separate discussion.)    < However, consider the memory capacities for the Alpha line:          (Max)  DS10   2 Gb      DS20E  4 Gb  ES40  16 Gb (model 1)        32 Gb (model 2)   E In other words, it isn't until you buy a system the size of an ES40,  F starting price around $30k (with Linux and 512 Mb memory) that you canJ begin to take advantage of this capability of the Alpha.  It's almost likeJ buying an airplane with no wings - you've _almost_ got everything you needH to fly, but without the wings none of the other flying related features K have any utility whatsoever.  If all you can do is taxi you'd be better off J driving a car.  And sticking with that image for minute, it's as if CompaqE is bragging about the avionics on this unflyable machine.  Here is an E example of Compaqs muddled, if not outrightly deceptive message, with  regard to this memory issue:    6    http://www5.compaq.com/alphaserver/ds20e/index.html  4 where you will find these two bullets in succession:     > up to 4 Gb ECC memory F   > Enhanced application support from 64-bit very large memory (VLM64)&     and advanced clustering technology  L Maybe I'm missing something but since the system cannot hold more than 4 Gb K of memory the VLM64 is entirely superfluous, right?   "VLM32" could just as 2 easily access every byte of memory in this system.  K Admittedly memory used to be prohibitively expensive, but 3rd party RAM for J a DS10 or DS20E now costs just over $1/Mb (Kingston, 512 Mb, $636 or $572,K respectively) and the general trend for memory prices is ever downward.  So H if these models had enough slots in them (or memory modules increased inJ density) adding 8Gb would only run about $8000. That's a lot of money, butF since the entry price for an ES40 is around $30K (with only 512Mb), itD would offer a very attractive way to obtain a large memory system.    B Anyway, my main point is that if Compaq is going to bother to makeK low end Alphas they should at least design them so that customers can take  I maximum advantage of the processor's capabilities.  But no low end alpha  J has ever been able to address >4Gb of memory.  That was understandable in J 1992, but it's 2001, and no longer makes sense.  I've often heard it said K that only only has to need the 33rd bit to drive an application to 64 bits. H But the flip side is that if all 64 bit systems are hideously expensive H there's a huge incentive to work around the 32 bit limit by other means.K Compaq needs to lower the energy barrier somewhat, so that people will see  I large memory applications as a real option, and then they can sell Alphas E (at a higher margin) in situations where they might otherwise have at C best sold a Proliant, and at worst, lost a sale to Dell, HP or IBM.    Regards,   David Mathog mathog@seqaxp.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech     ------------------------------  % Date: Fri, 26 Jan 2001 00:50:57 -0800 + From: "Wayne Holland" <wholland@tscnet.com> + Subject: A help message to Andrew Harrison. O Message-ID: <04648693C37BFF12.87115033C88B078C.4316F21E6790B777@lp.airnews.net>   
 Hi Andrew!  I      Seeing that you represent Sun Microsystems, can you help me with Sun  filters?B I have an Epson Stylus Color 600 printer... a very common printer.E I have up and running Solaris 8 x86.  This is a personal system.  The ! printer works in ascii text mode, F but using ghostscript is something else.  I've viewed the Solaris faqs followed the instructions.J The instructions didn't work.  All my printer does is spit out blank pages when I ask to print postscriptL documents.  In all honesty, Microsoft windows does a better job than Solaris in viewing web pagesI correctly and printing web pages correctly. What help can you offer me in  setting up my printer?  K My challenge to Sun.... make an operating system for the MASSES better than 
 Microsoft!   ------------------------------    Date: 26 Jan 2001 09:57:18 -0500/ From: jordan@lisa.gemair.com (Jordan Henderson) / Subject: Re: A help message to Andrew Harrison. ) Message-ID: <94s38e$es$1@lisa.gemair.com>   < As much as I've contributed to this newsgroup being for and > about Andrew Harrison, I apologize.  In most of those debates,> I can excuse myself as it being a matter of my making a point  of personal privilege.  < This is so far off topic for comp.os.vms.  Please take these8 kinds of questions to comp.os.solaris or comp.sys.sun.    = If you really want to discuss something like this with Andrew . Harrison and Andrew Harrison alone, try email.  O In article <04648693C37BFF12.87115033C88B078C.4316F21E6790B777@lp.airnews.net>, * Wayne Holland <wholland@tscnet.com> wrote: >Hi Andrew!  > J >     Seeing that you represent Sun Microsystems, can you help me with Sun	 >filters? C >I have an Epson Stylus Color 600 printer... a very common printer. F >I have up and running Solaris 8 x86.  This is a personal system.  The" >printer works in ascii text mode,G >but using ghostscript is something else.  I've viewed the Solaris faqs  >followed the instructions. K >The instructions didn't work.  All my printer does is spit out blank pages  >when I ask to print postscript M >documents.  In all honesty, Microsoft windows does a better job than Solaris  >in viewing web pages J >correctly and printing web pages correctly. What help can you offer me in >setting up my printer?  > L >My challenge to Sun.... make an operating system for the MASSES better than >Microsoft!  >  >  >    -Jordan Henderson  jordan@greenapple.   ------------------------------  % Date: Fri, 26 Jan 2001 12:18:41 -0800 ) From: Wayne Holland <wholland@tscnet.com> / Subject: Re: A help message to Andrew Harrison. O Message-ID: <74BA689A9095E311.D79D5C56C91E502A.0203B8C8F156F162@lp.airnews.net>    Jordan Henderson wrote:  > = > As much as I've contributed to this newsgroup being for and @ > about Andrew Harrison, I apologize.  In most of those debates,? > I can excuse myself as it being a matter of my making a point  > of personal privilege. > > > This is so far off topic for comp.os.vms.  Please take these8 > kinds of questions to comp.os.solaris or comp.sys.sun. > ? > If you really want to discuss something like this with Andrew 0 > Harrison and Andrew Harrison alone, try email. > Q > In article <04648693C37BFF12.87115033C88B078C.4316F21E6790B777@lp.airnews.net>, , > Wayne Holland <wholland@tscnet.com> wrote:
 > >Hi Andrew!  > > L > >     Seeing that you represent Sun Microsystems, can you help me with Sun > >filters? E > >I have an Epson Stylus Color 600 printer... a very common printer. H > >I have up and running Solaris 8 x86.  This is a personal system.  The$ > >printer works in ascii text mode,I > >but using ghostscript is something else.  I've viewed the Solaris faqs  > >followed the instructions. M > >The instructions didn't work.  All my printer does is spit out blank pages ! > >when I ask to print postscript O > >documents.  In all honesty, Microsoft windows does a better job than Solaris  > >in viewing web pages L > >correctly and printing web pages correctly. What help can you offer me in > >setting up my printer?  > > N > >My challenge to Sun.... make an operating system for the MASSES better than
 > >Microsoft!s > >s > >  > >  >  > -Jordan Henderson  > jordan@greenapple.  F I know it was off topic... but when you can't get any answers from the Sun newsgroupsG you start to look around for someone that may help.  Sorry if it is offP topic.   ------------------------------  # Date: Sat, 27 Jan 2001 01:32:52 GMTg' From: moi_is_me <moi_is_me@my-deja.com>e Subject: ACP-QIO and VBN) Message-ID: <94t8g4$s4k$1@nnrp1.deja.com>    Hi,L  /   Std Disclaimer ... please excuse my ignoranceL    B   I have been examing some code that uses ACP-QIO to create/extend   and write to disc.    =   However, there seems to be case where the code may write to07   a VBN outside of what has been allocated by the code.3  A   For example, if it has successfully allocated 4096 blocks,  ande4   then, in it's infinite wisdom, decides to write to?   VBN 5050 ... will the disc driver reject the request, or will6B   it silently accept it, with the possible consequence of trashing   some other file ?e   TIAr   Pierre     Sent via Deja.comn http://www.deja.com/   ------------------------------  % Date: Fri, 26 Jan 2001 23:05:16 -0500o2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: Backup Problem...7 Message-ID: <200101262305_MC2-C345-7DAC@compuserve.com>a  H         Not having support is sometimes a little awkward, isn't it?  :-)  H         Have you done an ANALYZE /DISK_STRUCTURE on the disk or disks in	 question?e  ! Message text written by Jim Agnewe >Dear all netbeings, =    J Hello and warm fuzzies and whatnot.  I've been seeing this error off and = on4 for a long time, and finally am getting tired of notG knowing what it's about exactly.  Since this is in a daily job, and our ; files turn over slowly, I've not really worried about this.s  H Can anyone enlighten me as to what backup-w-fidnotfnd really means? it's9 not mentioned in the system error message util except forsG "contact support", basically, and we don't have support.  oh, yeah, vmsr 5.5-2...  	 Jim Agnew     J %BACKUP-E-OPENIN, error opening ANACIN$DUA2:[000000]PD_FREE_FILE.DAT;1 as=  - input           -SYSTEM-W-NOSUCHFILE, no such J file                                                %BACKUP-I-STARTVERIFY= ,s starting verificationf* pass                                     =  6 %BACKUP-E-VERIFYERR, verification error for block 3 of( ANACIN$DUA2:[000000]000000.DIR;1       =  F %BACKUP-W-FIDNOTFND, internal error. File to be processed not found on process list  J %BACKUP-I-STARTRECORD, starting backup date recording pass               =   =t                     =o   $       SETUP ZIPe <f   ------------------------------  % Date: Fri, 26 Jan 2001 23:39:48 -0500f- From: JF Mezei <jfmezei.spamnot@videotron.ca>t Subject: Re: Backup Problem..., Message-ID: <3A72510F.361A49E1@videotron.ca>  K > %BACKUP-E-OPENIN, error opening ANACIN$DUA2:[000000]PD_FREE_FILE.DAT;1 asn/ > input           -SYSTEM-W-NOSUCHFILE, no suchu5 > file                                                  L I would suspect that PD_FREE_FILE.DAT;1  was deleted between the time BACKUPG built its file list and the time it got around to backing that file up.g  J When I backup my all mighty Microvax-II on a TK70, that process takes longP enough that there are quiet a few files (many .LOGs) that generate that message.  " > ANACIN$DUA2:[000000]000000.DIR;1H > %BACKUP-W-FIDNOTFND, internal error. File to be processed not found on > process list  L HELP/MESSAGE FIDNOTFND seems to imply you've hit a problem worthy of telling the VMS caretakers about it.   ------------------------------  % Date: Fri, 26 Jan 2001 11:26:40 +0000d% From: Alan Greig <agreig@my-deja.com>nV Subject: Compaq annual meeting of analysts and investors - VMS to feature allegedly...8 Message-ID: <91n27tkk7criogcf00e3lbhncd299r447n@4ax.com>  E Compaq's annual analysts meeting takes place at 9am EDT today (FridayaF 26th January). Information received says that VMS will feature in this@ presentation. Whether hastilty added after the full year results? conference tellingly spoke of UNIX, NT, Himalaya but not VMS orvD whether always planned it will be interesting to see just how VMS is/ pitched. WIll Michael Capellas say  the V word?u  D To view the conference you need Windows Media Player (I know). Go toE www.compaq.com and follow the link under news to register. You should - be able to replay it after the event as well.I     --
 Alan Greig   ------------------------------  % Date: Fri, 26 Jan 2001 18:25:02 +0000e0 From: andrew harrison <andrew.nospam@uk.sun.com>= Subject: Re: Compaq: A simple, affordable clustering solutionT* Message-ID: <3A71C0FE.2D0C6306@uk.sun.com>   "Main, Kerry" wrote: >  > Andrew ... > < > Again, give it a rest. We have Customers doing this today. >    Kerry when will you learn.  : Of course you can engineer a system that has zero downtime6 for example if you have to reboot a node in a cluster.  9 That isn't what I am objecting to and it wasn't what the a6 other people who disageed with you blanket claims were objecting ot either.  5 I have objected to your claim that this is a general n0 case, it isn't and that it it somehow unique to ! OpenVMS clusters, again it isn't.i  : And every time you and jlsue make this claim you reinforce2 just how little you appear to know about other non3 OpenVMS clusters and a whole group of applications g9 that do not conform to your general case and which cannotR5 survive the reboot of a node without interuption and s loss of state.  N > If you want to understand how Customers can shutdown individual servers in aH > OpenVMS cluster with zero application availability impact, please read > previous threads.s >   6 I have read the previous threads Kerry and nothing you2 have suggested is now a unique feature of OpenVMS.   Regardsf Andrew Harrisons Enterprise IT Architectd   ------------------------------  # Date: Fri, 26 Jan 2001 08:44:24 GMTm3 From: Piyush Avichal <pa@it.singer-friedlander.com>  Subject: CTRL Z from a scriptu) Message-ID: <94rdd7$7tt$1@nnrp1.deja.com>t  E Does anyone know how I can code in a CTRL-Z from within a DCL script?oD I am writing a script which requires a CTRL-Z to be sent to exit out
 of something.    TIAt   Piyush.t     Sent via Deja.comi http://www.deja.com/   ------------------------------  # Date: Fri, 26 Jan 2001 08:59:37 GMTe From: Didier.Morandi@gmx.fri! Subject: Re: CTRL Z from a scriptk) Message-ID: <94re9n$8f7$1@nnrp1.deja.com>o  ) In article <94rdd7$7tt$1@nnrp1.deja.com>,i6   Piyush Avichal <pa@it.singer-friedlander.com> wrote:G > Does anyone know how I can code in a CTRL-Z from within a DCL script?-F > I am writing a script which requires a CTRL-Z to be sent to exit out > of something.   
 $ ed a.com6 PF1 26 PF1 3 will insert a ^Z where the cursor is, so:   $ write out "^Z" $ exit  
 should do it.m   D.     Sent via Deja.com  http://www.deja.com/   ------------------------------  % Date: Fri, 26 Jan 2001 12:49:57 +0100 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>s! Subject: Re: CTRL Z from a scripto) Message-ID: <3A716465.EC7C017C@gtech.com>    Piyush Avichal wrote: G > Does anyone know how I can code in a CTRL-Z from within a DCL script?lF > I am writing a script which requires a CTRL-Z to be sent to exit out > of something.t  A Normally you use a line starting with a dollar in a DCL script to0
 indicate EOF.o   Example:   $ TYPE SYS$INPUT
 Hello world !  $b $ EXIT   Arne   ------------------------------  % Date: Fri, 26 Jan 2001 12:51:50 +0100e= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com> ! Subject: Re: CTRL Z from a scripto) Message-ID: <3A7164D6.F2B30048@gtech.com>n   Didier.Morandi@gmx.fr wrote:+ > In article <94rdd7$7tt$1@nnrp1.deja.com>, 8 >   Piyush Avichal <pa@it.singer-friedlander.com> wrote:I > > Does anyone know how I can code in a CTRL-Z from within a DCL script?lH > > I am writing a script which requires a CTRL-Z to be sent to exit out > > of something.  >  > $ ed a.com8 > PF1 26 PF1 3 will insert a ^Z where the cursor is, so: >  > $ write out "^Z" > $ exit >  > should do it.i  H Assuming that the script is some kind of client/server script that needs& to really send a ^Z to exit something.  E One could suspekt that the poster just want to exist a utility withinw theh DCL script.r   Arne   ------------------------------  % Date: Fri, 26 Jan 2001 12:52:44 +0100 = From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>l! Subject: Re: CTRL Z from a scripta) Message-ID: <3A71650C.4FC03AD0@gtech.com>a   Andy Burns wrote: 2 > On Fri, 26 Jan 2001 10:43:28 GMT, Piyush Avichal' > <pa@it.singer-friedlander.com> wrote:i+ > >So will this not work from a DCL Script:s > >  > >$ telnet xxx.xxx.xxx.xxxf > >login > >passwordr > >command1o > >command2  > >command3d > >...	 > >ctrl/zm > >exit( > >$ mail/subject .... > . > Try looking at the $DECK / $EOD commands ...  < Most TELNET implementation is not scriptable from a file, so! it wil not work DECK/EOD or not !L   Arne   ------------------------------    Date: 26 Jan 2001 10:21:49 +0100G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>i! Subject: Re: CTRL Z from a script H Message-ID: <y4puhamzwy.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  5 Piyush Avichal <pa@it.singer-friedlander.com> writes:E  G > Does anyone know how I can code in a CTRL-Z from within a DCL script?c  F CTRL-Z is the end-of-file signal. In a DCL command file, the next lineI starting with a $ tells DCL that this is the next command, and so it will D generate an end-of-file condition to the programm currently running.   Summary: You don't need that.e   	Jan   ------------------------------  # Date: Fri, 26 Jan 2001 10:43:28 GMTg3 From: Piyush Avichal <pa@it.singer-friedlander.com>a! Subject: Re: CTRL Z from a scriptf) Message-ID: <94rkcf$cr7$1@nnrp1.deja.com>m  ( So will this not work from a DCL Script:   $ telnet xxx.xxx.xxx.xxx login  password command1 command2 command3 ...r ctrl/z exit $ mail/subject ....i   Piyush    H In article <y4puhamzwy.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>,C   Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>c wrote:7 > Piyush Avichal <pa@it.singer-friedlander.com> writes:  > A > > Does anyone know how I can code in a CTRL-Z from within a DCL  script?u >rH > CTRL-Z is the end-of-file signal. In a DCL command file, the next lineF > starting with a $ tells DCL that this is the next command, and so it willF > generate an end-of-file condition to the programm currently running. >3 > Summary: You don't need that.  >e > 	Jan >o     Sent via Deja.com  http://www.deja.com/   ------------------------------  % Date: Fri, 26 Jan 2001 10:55:21 +0000)! From: Andy Burns <andy@burns.net>0! Subject: Re: CTRL Z from a scriptc8 Message-ID: <4rl27tsv320h2ddiof7r42sblhnakki8dv@4ax.com>  0 On Fri, 26 Jan 2001 10:43:28 GMT, Piyush Avichal% <pa@it.singer-friedlander.com> wrote:y  ) >So will this not work from a DCL Script:h >t >$ telnet xxx.xxx.xxx.xxx  >login	 >passwordi	 >command1t	 >command2r	 >command3  >... >ctrl/zt >exitt >$ mail/subject ....  , Try looking at the $DECK / $EOD commands ...   -- d
 Andy Burns   ------------------------------   Date: 26 Jan 2001 12:27:21 GMT3 From: gartmann@immunbio.mpg.de (Christoph Gartmann)o! Subject: Re: CTRL Z from a scripto0 Message-ID: <94rqf9$qav$1@n.ruf.uni-freiburg.de>  _ In article <94rkcf$cr7$1@nnrp1.deja.com>, Piyush Avichal <pa@it.singer-friedlander.com> writes: ) >So will this not work from a DCL Script:e >t >$ telnet xxx.xxx.xxx.xxxs >login	 >passwordo	 >command1e	 >command2e	 >command3a >....s >ctrl/zp >exitn >$ mail/subject ....  F This is a typical case for C-Kermit (http://www.columbia.edu/kermit/).   Regards,    Christoph Gartmannw  H -- --------------------------------------------------------------------+H | Max-Planck-Institut fuer      Phone   : +49-761-5108-464   Fax: -452 |H | Immunbiologie                                                        |H | Postfach 1169                 Internet: gartmann@immunbio.mpg.de     |H | D-79011  Freiburg, FRG                                               |H +--------- http://www.immunbio.mpg.de/home/english/menue.html ---------+   ------------------------------  + Date: Fri, 26 Jan 2001 08:34:22 +0000 (UTC) ' From: Osmo Kujala <kujala@tukki.jyu.fi>c" Subject: Re: dec 3000 300 crashing, Message-ID: <94rcqe$r6a$1@mordred.cc.jyu.fi>  3 Robert Deininger <rdeininger@mindspring.com> wrote:s  G > Let's take a poll.  How many folks with DEC 3000 300-ish systems passr >>>> T MEM LLSCt > and how many fail the test?r  I 3000/300lx, 2x32MB SIMMS (PC quality,not DEC): T MEM LLSC & T MEM both OK.  D Years a'go DEC maintenance did some "hardware patch", soldering someE component somewhere on main board of 3000/300LX (Dont remember was it G this same LX). (There were no problems before or after). I wonder mighta& that make a difference for mem tests. K Another difference might be number of SIMMs. I've tested many SIMMs for LX. G It should only take 32 and 8 MB (double banked?) SIMMs, but one pair of:J wrong SIMMs, say 16MB seems to work (maybe not reliably) but if more banks" are filled the type must be right.  H Hey, this memory system is funny. I tried two pairs of 32MB 36-bit SIMMsE and one pair of 4MB 33-bit SIMMs. Went through tests OK. Sho mem saysyJ four banks of 32MB and one bank of 8MB. Seems that first mem bank is buildH taking first halves of first pair of SIMMs, second bank is second halvesG of first pair etc. . So when there is pair of single banked SIMMs, thatdI makes only one memory bank and next mem bank is left empty. Because empty3J bank is not allowed in between, there can't be more SIMMs after the single banked pair.   regardst          Osmo Kujala   ------------------------------  % Date: Fri, 26 Jan 2001 04:30:07 -0500 2 From: rdeininger@mindspring.com (Robert Deininger)" Subject: Re: dec 3000 300 crashingL Message-ID: <rdeininger-2601010430070001@user-2ivebf4.dialup.mindspring.com>  p In article <wjs2-7E433B.14531425012001@newsstand.cit.cornell.edu>, "William J. Schaff" <wjs2@cornell.edu> wrote:  
 > In article iD > <rdeininger-2501011116440001@user-2iveal5.dialup.mindspring.com>, 5 > rdeininger@mindspring.com (Robert Deininger) wrote:  > I > > Let's take a poll.  How many folks with DEC 3000 300-ish systems passe > > >>> T MEM LLSC > > and how many fail the test?  >  > K > 3000-300 with 128M Kingston Memory, 32 Mb Dec - Fail - in a trouble free   > box (7 years?)     Well, I exercised our two boxes again today.  One failed all the memory tests until I did a console INIT.  Then it passed the whole thing, where it had failed on LLSC a few days ago.  4 The other one still failed on LLSC, even after INIT.   -- d Robert Deininger rdeininger@mindspring.com.   ------------------------------  % Date: Fri, 26 Jan 2001 04:31:19 -0500e2 From: rdeininger@mindspring.com (Robert Deininger)" Subject: Re: dec 3000 300 crashingL Message-ID: <rdeininger-2601010431200001@user-2ivebf4.dialup.mindspring.com>  W In article <3A70D518.5F1837AA@infopuls.com>, Christof Brass <brass@infopuls.com> wrote:p   > Robert Deininger wrote:aI > > Let's take a poll.  How many folks with DEC 3000 300-ish systems passf > > >>> T MEM LLSC > > and how many fail the test?A > >  > > -- > > Robert Deininger > > rdeininger@mindspring.coma > B > I can do the test but I have so many open X windows that I don't? > want to shutdown my system. It now up and running for 44 days A > and it may running another 44 days. If I have to shutdown or if ) > it crashes I'll do that test. Promised.n   LOL!  I know how that goes!  It takes me 10 minutes to close up all my windows when I have to shut down.  I hate it.  We had a scheduled power outage, so everything had to go.  What a pain.i   -- o Robert Deininger rdeininger@mindspring.com    ------------------------------  % Date: Fri, 26 Jan 2001 13:10:06 -0500i2 From: rdeininger@mindspring.com (Robert Deininger)" Subject: Re: dec 3000 300 crashingL Message-ID: <rdeininger-2601011310060001@user-2ive6bd.dialup.mindspring.com>  U In article <94rcqe$r6a$1@mordred.cc.jyu.fi>, Osmo Kujala <kujala@tukki.jyu.fi> wrote:n  5 > Robert Deininger <rdeininger@mindspring.com> wrote:n > I > > Let's take a poll.  How many folks with DEC 3000 300-ish systems pass  > >>>> T MEM LLSCJ > > and how many fail the test?p > K > 3000/300lx, 2x32MB SIMMS (PC quality,not DEC): T MEM LLSC & T MEM both OKu > F > Years a'go DEC maintenance did some "hardware patch", soldering someG > component somewhere on main board of 3000/300LX (Dont remember was it I > this same LX). (There were no problems before or after). I wonder mightQ( > that make a difference for mem tests. M > Another difference might be number of SIMMs. I've tested many SIMMs for LX. I > It should only take 32 and 8 MB (double banked?) SIMMs, but one pair oftL > wrong SIMMs, say 16MB seems to work (maybe not reliably) but if more banks$ > are filled the type must be right. > J > Hey, this memory system is funny. I tried two pairs of 32MB 36-bit SIMMsG > and one pair of 4MB 33-bit SIMMs. Went through tests OK. Sho mem says0L > four banks of 32MB and one bank of 8MB. Seems that first mem bank is buildJ > taking first halves of first pair of SIMMs, second bank is second halvesI > of first pair etc. . So when there is pair of single banked SIMMs, that-K > makes only one memory bank and next mem bank is left empty. Because emptyiL > bank is not allowed in between, there can't be more SIMMs after the single > banked pair.   Yes, this is documented (though some in obscure places).  The 300s have a very simple memory memory configuration system.  There's basically a single bit that says whether a bank is small (16 MB) or large (64 MB).  The manual says you have to put the larger SIMMs in first, and the suggested memory-finding algorithm assumes that.  That bit doesn't give a way to identify a 3rd size of SIMM.  The firmware could do it by trial and error, but the data structures used to tell the OS would have to change radically   And you are right, a "logical bank" is 1 side of 2 SIMMs, and the next logical bank is the other side of the same SIMMs.  A physical bank is a matched pair of adjacent SIMMs.  c Out-of-spec memory configurations might work, but they might break with other versions of firmware.C  pThe 3000 400 and higher systems have a much more flexible set of memory configuration registers, and can handle more sizes. I've heard of folks putting the memory risers from a 500/800/900 into a 400/600/700 system (the lid won't shut) and thereby doubling the memory capacity.  That probably means the firmware is the same in all the systems wrt memory configuration.  , None of this explains the LLSC test, though.   --   Robert Deininger rdeininger@mindspring.come   ------------------------------   Date: 26 Jan 2001 14:07:43 GMT1 From: bill@triangle.cs.uofs.edu (Bill Gunshannon) . Subject: Re: DEC TCPIP V5.0 problem with Linux, Message-ID: <94s0bf$2un8$1@info.cs.uofs.edu>  E If I had to make a quick guess, from your description I would suspect E that one of the NICs (most likely the Linux PC) is set to full duplexTG and hooked up to a half duplex hub.  That's just a guess, but it's what F I would look at first.  Question?  Is the NIC in the Linux Box a newerH one using the RealTEK chipset??  I had a lot of trouble using these withF either Linux or FreeBSD.  The primary visible symptom was an extremely high collision rate.  $ Good luck tracking the problem down.   bill   -- uJ Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   o   ------------------------------  # Date: Fri, 26 Jan 2001 09:32:35 GMT- From: Didier.Morandi@gmx.fr ) Subject: difference between EV6 and EV67?<) Message-ID: <94rg7k$9qc$1@nnrp1.deja.com>l  7 Why Rdb 7.0.4 doesn't run with an EV67 Alpha processor?l  B My question focuses more on the microcode reason than on other RdbH related stuff. Is it because Rdb 7.0.4 uses a microcode instruction that, is no more implemented in the EV67, or what?  D Generally speaking, what is the technical reason for which a program9 doesn't run on a particular new version of an Alpha chip?a   Thanks,V   D.     Sent via Deja.comd http://www.deja.com/   ------------------------------    Date: 26 Jan 2001 16:08:24 +0100G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de> - Subject: Re: difference between EV6 and EV67?wH Message-ID: <y4snm62vx3.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  N I thought it was EV6 processors in general? The EV6 actually makes use of someM of the more relaxed memory consistency features of the Alpha architecture, asCL I understand it, and previous compilers generated broken LL/SC code that wasH OK on EV4 and EV5, but could break on EV6 systems. DEC^H^H^HCompaq has aL program to go through executables and find candidates for such mistaken codeF sequences, but it isn't fool proof (it can generate false positives). J Apparently, Oracle wasn't prepared to check and/or re-issue older versions2 of its two VMS database products for this feature.   	Jan   ------------------------------  % Date: Fri, 26 Jan 2001 15:40:14 +0000o- From: Tim Llewellyn <tim.llewellyn@bbc.co.uk>i- Subject: Re: difference between EV6 and EV67? ) Message-ID: <3A719A5E.D6113759@bbc.co.uk>e   Jan Vorbrueggen wrote:  P > I thought it was EV6 processors in general? The EV6 actually makes use of someO > of the more relaxed memory consistency features of the Alpha architecture, aseN > I understand it, and previous compilers generated broken LL/SC code that wasJ > OK on EV4 and EV5, but could break on EV6 systems. DEC^H^H^HCompaq has aN > program to go through executables and find candidates for such mistaken codeG > sequences, but it isn't fool proof (it can generate false positives).KL > Apparently, Oracle wasn't prepared to check and/or re-issue older versions4 > of its two VMS database products for this feature. >-  K So, Compaq broke VMS's backwards compatibilty, one of its' major strengths.h+ No wonder Oracle ran away again eventually.j   >C
 >         Janr   --6 Tim Llewellyn, OpenVMS Infrastructure, Remarcs Project0 MedAS at the BBC, Whiteladies Road, Bristol, UK.A Email tim.llewellyn@bbc.co.uk. Home tim.llewellyn@cableinet.co.ukt  A I speak for myself only and my views in no way represent those ofo MedAS or the BBC..   ------------------------------    Date: 26 Jan 2001 07:11:45 -07001 From: nothome@spammers.are.scum (Malcolm Dunnett):- Subject: Re: difference between EV6 and EV67?$, Message-ID: <vLM0sF4qwllI@malvm1.mala.bc.ca>  * In article <3A719A5E.D6113759@bbc.co.uk>, 3     Tim Llewellyn <tim.llewellyn@bbc.co.uk> writes:I  M >> Apparently, Oracle wasn't prepared to check and/or re-issue older versionse5 >> of its two VMS database products for this feature.  >> > M > So, Compaq broke VMS's backwards compatibilty, one of its' major strengths.^  N     I don't think I'd call this breaking backwards compatibility. The programsL will run just fine under VMS 7.2 on processors older than EV6. It seems thatL what happened was that Compaq found a bug ( or misfeature ) in the compilersM which was generating code that didn't conform to the Alpha architecture. TheylK fixed this bug for things to work right on EV6, and provided a tool to helpiN users identify executables which contained the bad code. If one has the sourceL code all that would be required to fix things would be a recompile/link. I'mI guessing that Oracle still has the source code for those earlier versions  of their products.  M I thought it was rather interesting that Oracle chose to add a check into RdbeJ which prevented it from running under VMS 7.2, purportedly because of thisH EV6 problem. A more reasonable approach might have been to check for theI processor type and decide based on that. A cynic would point out that theiM "version check" thingy has the added benefit of forcing people to upgrade Rdb I if they upgrade to a newer VMS version, thus providing more incentive foru2 customers to keep paying maintenance fees for Rdb.   ------------------------------  % Date: Fri, 26 Jan 2001 16:25:45 +0000,% From: Alan Greig <agreig@my-deja.com> - Subject: Re: difference between EV6 and EV67?t8 Message-ID: <a2937tgbe3lfemngn7mgoi2f4q1rflkr3f@4ax.com>  1 On Fri, 26 Jan 2001 15:40:14 +0000, Tim Llewellynh  <tim.llewellyn@bbc.co.uk> wrote:   >i >w >Jan Vorbrueggen wrote:r > Q >> I thought it was EV6 processors in general? The EV6 actually makes use of somerP >> of the more relaxed memory consistency features of the Alpha architecture, asO >> I understand it, and previous compilers generated broken LL/SC code that was K >> OK on EV4 and EV5, but could break on EV6 systems. DEC^H^H^HCompaq has aeO >> program to go through executables and find candidates for such mistaken codexH >> sequences, but it isn't fool proof (it can generate false positives).M >> Apparently, Oracle wasn't prepared to check and/or re-issue older versions 5 >> of its two VMS database products for this feature.n >> >iL >So, Compaq broke VMS's backwards compatibilty, one of its' major strengths., >No wonder Oracle ran away again eventually.  C No, not really. There was a bug in the earlier version of the BlissaF compiler used by Oracle which generated bad code. The code worked fineF on procesors prior to EV4 but it violated some rules even on EV4. Only? on EV6 did later optimizations of the processor cause problems.   F This was the offical reason given for dropping RDB on Alpha/NT because( Compaq fixed Bliss on VMS but not for NT   >> >>         Jan   --
 Alan Greig   ------------------------------  % Date: Fri, 26 Jan 2001 12:31:12 -0500l5 From: "Fred Kleinsorge" <kleinsorge@star.zko.dec.com>r- Subject: Re: difference between EV6 and EV67?-, Message-ID: <94sc8r$9uhh$1@lead.zk3.dec.com>  E Correct.  SRM_CHECK is the utility, and it can be found in SYS$SYSTEMc  C The problem is a bug in the code generation for old versions of them compilers, especiallyr= BLISS and MACRO-32, but also in some "exotic" C code as well.o  L The issue had to do with what can be done between a load locked, and a storeG conditional.  On older CPU's, it turns out quite a lot - as long as youcJ didn't touch data within a certain distance from the lock (usually a cacheK line), they also had a global "lock" state that was maintained.  On EV6 and G beyond, they used a different technique for implementing the lock using H cache state and a local (not global) locked flag.  So this marginal codeG sequence (if you read the SRM closely it discouraged it strongly) wouldt always fail on EV6.   K Most of the problem had been fixed long ago (before it ever was a problem),nJ but you would be suprised at how many applications and 3rd party librariesH have not been recompiled since, say 1992, or had their compilers updated (especially BLISS).<       _Fredt    $ Malcolm Dunnett wrote in message ...* >In article <3A719A5E.D6113759@bbc.co.uk>,4 >    Tim Llewellyn <tim.llewellyn@bbc.co.uk> writes: >iE >>> Apparently, Oracle wasn't prepared to check and/or re-issue olderf versions6 >>> of its two VMS database products for this feature. >>>  >>C >> So, Compaq broke VMS's backwards compatibilty, one of its' majorc
 strengths. >oF >    I don't think I'd call this breaking backwards compatibility. The programsH >will run just fine under VMS 7.2 on processors older than EV6. It seems thatC >what happened was that Compaq found a bug ( or misfeature ) in the9	 compilers@I >which was generating code that didn't conform to the Alpha architecture.A TheyL >fixed this bug for things to work right on EV6, and provided a tool to helpH >users identify executables which contained the bad code. If one has the sourceI >code all that would be required to fix things would be a recompile/link.y I'm J >guessing that Oracle still has the source code for those earlier versions >of their products.- >-J >I thought it was rather interesting that Oracle chose to add a check into RdbiK >which prevented it from running under VMS 7.2, purportedly because of this I >EV6 problem. A more reasonable approach might have been to check for the J >processor type and decide based on that. A cynic would point out that theJ >"version check" thingy has the added benefit of forcing people to upgrade RdbgJ >if they upgrade to a newer VMS version, thus providing more incentive for3 >customers to keep paying maintenance fees for Rdb.s >K   ------------------------------  % Date: Fri, 26 Jan 2001 14:32:23 -0800D! From: Shane.F.Smith@Healthnet.com - Subject: Re: difference between EV6 and EV67?wD Message-ID: <OFD61589CD.B1EAF544-ON882569E0.007BA623@foundation.com>  I You could see it that way. I see it as the chip guys enforced some of the I original rules documented in the architecture manuals, and all the peopleoI who'd broken those rules got bit, including some VMS bits. It's like whenhH people used to play fast and loose with the VMS calling standard on VAX,' then got bit when the Alpha came along.w   Shaneh          A Tim Llewellyn <tim.llewellyn@bbc.co.uk> on 01/26/2001 07:40:14 AMr  ) Please respond to tim.llewellyn@bbc.co.ukt   To:   Info-VAX@Mvb.Saic.Comd cc:   . Subject:  Re: difference between EV6 and EV67?         Jan Vorbrueggen wrote:  K > I thought it was EV6 processors in general? The EV6 actually makes use ofo some> > of the more relaxed memory consistency features of the Alpha architecture, asJ > I understand it, and previous compilers generated broken LL/SC code that wasBJ > OK on EV4 and EV5, but could break on EV6 systems. DEC^H^H^HCompaq has aI > program to go through executables and find candidates for such mistakena codeG > sequences, but it isn't fool proof (it can generate false positives).iC > Apparently, Oracle wasn't prepared to check and/or re-issue oldera versions4 > of its two VMS database products for this feature. >   K So, Compaq broke VMS's backwards compatibilty, one of its' major strengths.0+ No wonder Oracle ran away again eventually.R   > 
 >         Janj   --6 Tim Llewellyn, OpenVMS Infrastructure, Remarcs Project0 MedAS at the BBC, Whiteladies Road, Bristol, UK.A Email tim.llewellyn@bbc.co.uk. Home tim.llewellyn@cableinet.co.uk"  A I speak for myself only and my views in no way represent those ofe MedAS or the BBC.4   ------------------------------  # Date: Fri, 26 Jan 2001 13:28:40 GMTR/ From: "John Nixon" <jorlnixon@worldnet.att.net>l( Subject: Examine log name on other node?D Message-ID: <cYec6.520$mA1.54190@bgtnsc06-news.ops.worldnet.att.net>  G In either VMS 7.1 or 7.2-1, is it possible to examine a logical name on E another node from within a command procedure, other than with SYSMAN?a   ------------------------------  # Date: Fri, 26 Jan 2001 14:39:40 GMTp/ From: "John Nixon" <jorlnixon@worldnet.att.net>C, Subject: Re: Examine log name on other node?F Message-ID: <M_fc6.9709$7b2.661814@bgtnsc05-news.ops.worldnet.att.net>  J I posted the previous question based on an E-Mail query that I got.  I now+ have a better idea of what they want to do.r  L They want to be able to tell if another node, in the same cluster, is up andI if the application is available so they can send a remote command to that3I node.  Right now, they send the command blindly, and if the node is down,D/ they are not aware the command did not execute.y  J Instead of asking how to perform a specific task, we should have asked for* suggestions on how to accomplish our goal.   Thanks,     Johni  : "John Nixon" <jorlnixon@worldnet.att.net> wrote in message> news:cYec6.520$mA1.54190@bgtnsc06-news.ops.worldnet.att.net...I > In either VMS 7.1 or 7.2-1, is it possible to examine a logical name onIG > another node from within a command procedure, other than with SYSMAN?O >  >n   ------------------------------  # Date: Fri, 26 Jan 2001 14:45:37 GMT = From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)g, Subject: Re: Examine log name on other node?0 Message-ID: <009F6B05.32FAC776@SendSpamHere.ORG>  v In article <cYec6.520$mA1.54190@bgtnsc06-news.ops.worldnet.att.net>, "John Nixon" <jorlnixon@worldnet.att.net> writes:H >In either VMS 7.1 or 7.2-1, is it possible to examine a logical name onF >another node from within a command procedure, other than with SYSMAN? >m >r  9 Sure.  Put the logical in the cluster-wide logical space.    --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMa             O city, n., 1. a place where trees are cut down and streets are named after them.t   ------------------------------  % Date: Fri, 26 Jan 2001 15:17:17 +0000 - From: Tim Llewellyn <tim.llewellyn@bbc.co.uk>s, Subject: Re: Examine log name on other node?) Message-ID: <3A7194FD.7BC993F9@bbc.co.uk>D   John Nixon wrote:   L > I posted the previous question based on an E-Mail query that I got.  I now- > have a better idea of what they want to do.  >fN > They want to be able to tell if another node, in the same cluster, is up andK > if the application is available so they can send a remote command to thatsK > node.  Right now, they send the command blindly, and if the node is down,e1 > they are not aware the command did not execute.  >f  N So, the cleanest way to me seems to ensure that the program sending the remoteP command returns a decent condition code ("node unavailable") if it fails to send theh command.  M Of course, if its a crap application that you have no control over, you couldc try pinging M the other node first, or opeining a decnet link if you are running decnet ande checking the status return.  M You can build bulletproof code with apps that don't return decent status, you 	 just haver to work a lot harder.e   >aL > Instead of asking how to perform a specific task, we should have asked for, > suggestions on how to accomplish our goal. > 	 > Thanks,e	 >    Johnl >a< > "John Nixon" <jorlnixon@worldnet.att.net> wrote in message@ > news:cYec6.520$mA1.54190@bgtnsc06-news.ops.worldnet.att.net...K > > In either VMS 7.1 or 7.2-1, is it possible to examine a logical name ontI > > another node from within a command procedure, other than with SYSMAN?i > >d > >r   --6 Tim Llewellyn, OpenVMS Infrastructure, Remarcs Project0 MedAS at the BBC, Whiteladies Road, Bristol, UK.A Email tim.llewellyn@bbc.co.uk. Home tim.llewellyn@cableinet.co.ukd  A I speak for myself only and my views in no way represent those ofI MedAS or the BBC.    ------------------------------  # Date: Fri, 26 Jan 2001 15:50:53 GMTs* From: Alan E. Feldman <alan48@my-deja.com>, Subject: Re: Examine log name on other node?) Message-ID: <94s6co$s4u$1@nnrp1.deja.com>e  F In article <M_fc6.9709$7b2.661814@bgtnsc05-news.ops.worldnet.att.net>,2   "John Nixon" <jorlnixon@worldnet.att.net> wrote:E > I posted the previous question based on an E-Mail query that I got.s I nowm- > have a better idea of what they want to do.  >oG > They want to be able to tell if another node, in the same cluster, isi up andF > if the application is available so they can send a remote command to thatE > node.  Right now, they send the command blindly, and if the node isD down,P1 > they are not aware the command did not execute.1 >fB > Instead of asking how to perform a specific task, we should have	 asked for , > suggestions on how to accomplish our goal.  ? Precisely!!! If only more posters would realize this. Anyway...h   I use'   $!
 $    SET NOONt $    SET VERIFY ) $    MCR NCP TELL 'NODE' SHOW EXEC TO NL:e& $    STATUS = $STATUS ! 'F$VERIFY(P3)'  $    ON WARNING THEN GOTO _ERROR $    SH SYM STATUS8 $    IF (STATUS .EQS. "%X00048084") THEN GOTO _DEL_ENTRY2 $    IF (.NOT. STATUS) THEN GOTO _NODE_UNREACHABLE  E in one of my routines, but we have no clusters. But this should work.cF Someone else may be able to suggest an easier way that works only in aA cluster. If I think of something better later, I'll post it, time  permitting.n  G Of course you must put the node name in the symbol node, and the verify-E comment is not needed. The ON WARNING statement should be replaced byi> the error handling you had in effect before the SET NOON or to0 something appropriate to your command procedure. --F NOTE: If you wish to e-mail me, please do NOT use the deja address. ItD is not a valid address. Instead, use the address below, removing the long wrong part first. Thanks.o   Disclaimer: JMHO Alan E. Feldman  &-)( afeldman@gfigroup.ButItSaidItPrinted.com     Sent via Deja.comr http://www.deja.com/   ------------------------------   Date: 26 Jan 2001 09:43 CSTV' From: carl@gerg.tamu.edu (Carl Perkins)w, Subject: Re: Examine log name on other node?- Message-ID: <26JAN200109435783@gerg.tamu.edu>o  3 "John Nixon" <jorlnixon@worldnet.att.net> writes....K }I posted the previous question based on an E-Mail query that I got.  I now., }have a better idea of what they want to do. } M }They want to be able to tell if another node, in the same cluster, is up andeJ }if the application is available so they can send a remote command to thatJ }node.  Right now, they send the command blindly, and if the node is down,0 }they are not aware the command did not execute. } K }Instead of asking how to perform a specific task, we should have asked fora+ }suggestions on how to accomplish our goal.  }  }Thanks, }   John  D GETSYI, in either the LIB$ form or the SYS$ form. You should be able- to use it to find out of a node is up or not.S  H As for "if the aplication is available", how to do that probably dependsJ on what you mean. If it is a process that should be running, you can checkH for it using GETJPI (again in either the LIB$ or SYS$ form, depending on your needs).   --- Carl   ------------------------------  # Date: Fri, 26 Jan 2001 16:09:16 GMTn* From: Alan E. Feldman <alan48@my-deja.com>, Subject: Re: Examine log name on other node?) Message-ID: <94s7f3$t6i$1@nnrp1.deja.com>n  ) In article <94s6co$s4u$1@nnrp1.deja.com>, -   Alan E. Feldman <alan48@my-deja.com> wrote:,H > In article <M_fc6.9709$7b2.661814@bgtnsc05-news.ops.worldnet.att.net>,4 >   "John Nixon" <jorlnixon@worldnet.att.net> wrote:G > > I posted the previous question based on an E-Mail query that I got.i > I now / > > have a better idea of what they want to do.r > >sF > > They want to be able to tell if another node, in the same cluster, is > up andH > > if the application is available so they can send a remote command to > thatG > > node.  Right now, they send the command blindly, and if the node isl > down,V3 > > they are not aware the command did not execute.  > >lD > > Instead of asking how to perform a specific task, we should have > asked fore. > > suggestions on how to accomplish our goal. > A > Precisely!!! If only more posters would realize this. Anyway...- >- > I use  >a > $! > $    SET NOON5 > $    SET VERIFYm+ > $    MCR NCP TELL 'NODE' SHOW EXEC TO NL:c( > $    STATUS = $STATUS ! 'F$VERIFY(P3)'" > $    ON WARNING THEN GOTO _ERROR > $    SH SYM STATUS: > $    IF (STATUS .EQS. "%X00048084") THEN GOTO _DEL_ENTRY4 > $    IF (.NOT. STATUS) THEN GOTO _NODE_UNREACHABLE >tG > in one of my routines, but we have no clusters. But this should work.aH > Someone else may be able to suggest an easier way that works only in aC > cluster. If I think of something better later, I'll post it, time'
 > permitting.a >gB > Of course you must put the node name in the symbol node, and the verifyG > comment is not needed. The ON WARNING statement should be replaced byo@ > the error handling you had in effect before the SET NOON or to2 > something appropriate to your command procedure.  2 Actually, you can probably just use something like  , $ IF (F$GETSYI("CLUSTER_MEMBER","nodename")) $ THEN$ $     !node is up and in the cluster $ ELSE* $     !node is not a member of the cluster $ ENDIFg  C but if the node is up and not in the cluster, the above will not beu able to tell you if it is up.i  F Is there some reason "they" cannot log in to the node on which the appF is running? If so, the NCP solution I originally suggested is probably7 the best, since you really only care if the node is up.    --F NOTE: If you wish to e-mail me, please do NOT use the deja address. ItD is not a valid address. Instead, use the address below, removing the long wrong part first. Thanks.c   Disclaimer: JMHO Alan E. Feldman  &-)( afeldman@gfigroup.ButItSaidItPrinted.com     Sent via Deja.comt http://www.deja.com/   ------------------------------  % Date: Fri, 26 Jan 2001 14:45:07 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca>e, Subject: Re: Examine log name on other node?, Message-ID: <3A71D3C2.BAE2C115@videotron.ca>   John Nixon wrote:@ > I > In either VMS 7.1 or 7.2-1, is it possible to examine a logical name onbG > another node from within a command procedure, other than with SYSMAN?   J You could use a DECNET task to do that. Look for TELL.COM in freeware CD.   0 However, with 7.2, you have clusterwide logicals on on node VELO:  ; DEFINE/TABLE=LNM$SYSCLUSTER_TABLE prefered_cake "CHOCOLATE"L    M This logical magically becomes available as a local logical name on all nodesCN in the cluster within a few micro seconds. Updates are possible to it as well.   ------------------------------  % Date: Fri, 26 Jan 2001 14:55:03 -0500e- From: JF Mezei <jfmezei.spamnot@videotron.ca>a, Subject: Re: Examine log name on other node?, Message-ID: <3A71D614.903CF692@videotron.ca>  2 re: finding if application is alive on other node:  L If it is a cluster, the cleanest thing to do would be for the application toN take a lock, and for the tester to try to take the lock. If it fails, it meansJ that the application is still alive and hoarding its lock. If the teste is= able to take the lock, it means that the application is dead.n  L Another way is to have the application create a mailbox that accepts controlM messages (one of which would be "are you there ?". Then you can use decnet toiN connect to that mailbox, send a "are you there" message and wait for an answerN (with a timeout, which when it expires means your application is not healthy).   As variations to this:M the application can create a decnet object. This object will disapear as soon$9 as the application dies. So a mere "OPEN/READ/WRITE TEMP iK "NODE::TASK=CHOCOLATE"  would do the trick since the open would fail if thenN application isn't there. (You'd still have to havce code to accept the call at the other end though).  . You could also use ICC services or even TCPIP.   ------------------------------    Date: 26 Jan 2001 09:54:05 +0100G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>h% Subject: Re: fast_io_copy.c questionsuH Message-ID: <y4snm6n176.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  4 mathog@seqaxp.bio.caltech.edu (David Mathog) writes:  K > Apparently this is because the program is hardwired to use buffers=16 and J > blocks=127.  Do the IO_* system calls interact with SET RMS/block/buffer@ > or must these values be hardwired for each IO_* based program?  M You have to code them explicitly. Of course, you could read these settings assG RMS does. IIRC, the system wide settings are SYSGEN parameters, and theeI process wide ones are in P1 space somewhere (look for an appropriate CTL$) symbol in the map file).   	Jan   ------------------------------    Date: 26 Jan 2001 09:51:19 +0100G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de> % Subject: Re: fast_io_copy.c questionsrH Message-ID: <y4vgr2n1bs.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  4 mathog@seqaxp.bio.caltech.edu (David Mathog) writes:  I > I guess .003 seconds was out of the question.  The test file was 16262 h( > blocks and it took .008 seconds to do: >  > $ copy infile outfile  > J > with everything on the RAMdisk, but to read all that data and then write > it back out again is:o > - >  2*16262*512/.008 = 2081536000 bytes/secondr > L > or 2Gb/sec.  That's the full memory bandwidth (more or less) on a DS10, so+ > it just cannot be done faster than that.    , So they really optimized the RAMdisk driver!  N Is your next conclusion the writers of the CRTL should please convert to usingN FAST_IO (with the current limitations going away, as per Fred via Rob) and useM RMS only for file name parsing et al.? Then you'd finally be as fast as Linuxg3 (because you're at the speed of light already 8-)).   N > Since FAST_IO_COPY takes almost exactly twice as long it must have an extra M > copy in it somewhere that's missing in COPY.  Maybe FAST_IO_COPY is moving c0 > the data from a read buffer to a write buffer?    Seems a likely good explanation.   	Jan   ------------------------------  # Date: Fri, 26 Jan 2001 21:50:31 GMTg- From: "Dave Pampreen" <davepampreen@home.com> ) Subject: Re: Filename wildcards under VMS > Message-ID: <Himc6.255751$hD4.61840910@news1.rdc1.mi.home.com>  F If yo are truly looking for portability, then you need to restate your request:  K The issue is NOT with C, rather the OS (VMS).  I suggest that you port your.I program without parsing the wildcards, and put the parsing in a DCL file.I  6 $ argc = "''P1'"     ! User passes a wildcard filename $ new_argc = ""n $e8 $ loop:    ! Parse 'argc' and get all matching filenames $ filename = f$search (argc)% $ if filename .eqs. "" then goto doneu& $ new_argc = "''new_argc' ''filename'" $ goto loopa $o $ done:h! $ yourprogram == "$<dir>progname"g) $ define /user_mode sys$input sys$commandl $ yourprogram 'new_argc' $ exit  C This way the program is portable, and the script is specific to VMSo   That's my $0.02  Dave    < "Nigel Arnot" <sysmgr@maxwell.ph.kcl.ac.uk> wrote in message1 news:009F6B14.D4EF5175.17@maxwell.ph.kcl.ac.uk... L > > >     I have got some utilities written in C that I'm porting to VMS and InJ > > > am having trouble with wildcards.  Under Unix a command like "myprogG > > > *.*" has the wildcard expanded by the shell and the filenames areV passedJ > > > to the utility via argv.  In VMS this does not happen.  Any thought? > >b > I > The runtime library LIB$FIND_FILE will expand a wildcard, returning onel > filename per call. >a2 > It ought to be quite easy to implement a routine >t6 > char **vmsargv( char** argv, int argc, int* vmsargc) >dL > which takes argc and argv from main() and builds up vmsargc and vmsargv byH > copying arg strings that do not contain the character '*' not prefixedH > by "\" and enumerating those that do with lib$find_file (and malloc toJ > create somewhere to put the results as a c-style string). Having written > this you could simply replace  >.  > main ( int argc, char ** argv) >  > by >f$ > main ( int vargc, char ** vargv) { >   ...e* >   argv = vmsargv ( vargv, vargc, &argc); >- >-G > and the rest of the unix-style code should then run as if it had been@H > invoked from a unix shell with no code changes.  Maybe a bit more work than9 > ad-hoc  expansion, but you  only have to write it once.t > J > (I'd be surprised if someone hasn't already done this, but it's not me.) >o > Yours,
 > Nigel Arnotu > NRA@MAXWELL.PH.KCL.AC.UK > 7 > "In the beginning there was nothing, which exploded."1   ------------------------------  % Date: Fri, 26 Jan 2001 23:36:28 -0000:= From: "Malcolm MacArthur" <malcolmm@rustic-place.demon.co.uk>6) Subject: Re: Filename wildcards under VMS A Message-ID: <980551652.18624.0.nnrp-12.d4e4c55f@news.demon.co.uk>   " Dave Pampreen wrote in message ...G >If yo are truly looking for portability, then you need to restate yourm	 >request:i >pL >The issue is NOT with C, rather the OS (VMS).  I suggest that you port yourJ >program without parsing the wildcards, and put the parsing in a DCL file. > 7 >$ argc = "''P1'"     ! User passes a wildcard filenamea >$ new_argc = "" >$9 >$ loop:    ! Parse 'argc' and get all matching filenames  >$ filename = f$search (argc) & >$ if filename .eqs. "" then goto done' >$ new_argc = "''new_argc' ''filename'"e >$ goto loop >$ >$ done:" >$ yourprogram == "$<dir>progname"* >$ define /user_mode sys$input sys$command >$ yourprogram 'new_argc'e >$ exitn >eD >This way the program is portable, and the script is specific to VMS >  >That's my $0.02 >Dave-J But you will then be limited by DCL's substitution linit of 255 charactersD for a symbol (A symbol can be larger... but you would still run intoK problems). Even at the maximum limit (1024? 2048? One of the two anyway...IYK can't check just now), expanding a wildcard for several hundred files wouldn fail the limit.>  I I ran into this problem recently when trying to get one program to run onmL another Alpha using task-to-task communication. DCL could not handle symbols0 of the size I needed for the whole command line.  I Eventually I ended up having to write the whole thing using CLI$ routines  ;-)X  	 -Malcolm.(   ------------------------------  % Date: Fri, 26 Jan 2001 23:04:58 -0500t2 From: "Richard B. Gilbert" <DRAGON@compuserve.com>% Subject: Filename wildcards under VMS:7 Message-ID: <200101262305_MC2-C345-7DAA@compuserve.com>k  $ Message text written by Bruce Pagram >HidH     I have got some utilities written in C that I'm porting to VMS and ID am having trouble with wildcards.  Under Unix a command like "myprogH *.*" has the wildcard expanded by the shell and the filenames are passedD to the utility via argv.  In VMS this does not happen.  Any thought?   Thanks in advance<  H         Well, you could do it the VMS way and expand the wildcarded fileJ name(s) in the program or you could get a copy of Jym Dyer's "shell_mung"=  J which will replace the original command line with one that looks as if it=  G had been processed by the shell.  The wild card expansion process is, I E believe, identical; the difference is how the expanded file names areiJ handled.  In the one case you expand the wildcarded filespec once for eac= h J matching file and process it, in the other you make a list, a la shell, a= nd! then process files from the list.o  B         There are advantages to theVMS way.  See "The Unix Hater's
 Handbook".  5         Shell_mung is distributed with Gnu grep v2.0.    ------------------------------  % Date: Fri, 26 Jan 2001 14:22:26 -0800h0 From: "Cokely, Jay" <COKELJA@mail.northgrum.com> Subject: FTP to GDG failso? Message-ID: <1D594B50D578D1119F9000805F150E770558012D@XCGCA017>5  K We are having problems FTP'ing from a DEC machine to a GDG (Generation dataiH set) in an IBM/MVS environment.  FTP's to non GDG names work fine.   TheG script below works from other environments but fails with the followingy message on the DEC processor.  s  < 	554 Requested action not taken: GDG name conversion failed.     The script used follows:L **************************************************************************** $set default sys$login1 $ftp /user=$XYZS02W /pass=??????? n1etso.grum.com- asciiaL quote site filetype=seq cylinders primary=50 secondary=25 recfm=fb lrecl=869' put dwhead.txt 'ISDW.ROSS.HEADERIN(+1)'n exitL ****************************************************************************. **********************************************  J Any ideas as to why this doesn't work, or where I might find documentation of the "554" error?r     Thanks,s     Jay Cokely  0655/32  PHONE (310) 332-3977 FAX      (310) 263-5303    ------------------------------  % Date: Fri, 26 Jan 2001 19:25:54 -0500t- From: JF Mezei <jfmezei.spamnot@videotron.ca>r Subject: Re: FTP to GDG fails , Message-ID: <3A721591.76363FEC@videotron.ca>   "Cokely, Jay" wrote:3 > $ftp /user=$XYZS02W /pass=??????? n1etso.grum.com  > asciitN > quote site filetype=seq cylinders primary=50 secondary=25 recfm=fb lrecl=869) > put dwhead.txt 'ISDW.ROSS.HEADERIN(+1)'r  J are you sure that you need to specify the file attributes for a generationL dataset ? Isn't this just a glorified partitioned dataset where you create a' member inside a pre-allocated dataset ?   N Also, I can't remember the exact nomenclature for generation datasets, but areM you sure that the (+1) is the right way to specify it ? As I recall, there isEM a way to specify a dataset so that a new version is created automatically, aseN well as specifying one where the system gives you the latest without having to# know what is the generation number.l  J I remember doing work with the SNA-DTF product where the VAX would get theK generation number of the dataset it was about to read and log it on the vax I and verify that it matched the generation number entered by the operator.   J Also, it is possible that the dataset itself has runned out of the controlM blocks (is that the term) which are needed for partitioned datasets. You needeE to do the equivalent of the re-organise on the mainframe to fix that.t  / These are just ideas. It has been too long ago.)   ------------------------------  % Date: Fri, 26 Jan 2001 14:43:20 +0100w= From: Arne =?iso-8859-1?Q?Vajh=F8j?= <arne.vajhoej@gtech.com>2- Subject: Re: ftp'ing savesets from W2K to VMSP) Message-ID: <3A717EF8.F31F7EE9@gtech.com>o   Tom Linden wrote:DJ > I think one needs to change block size.  Can someone give me the command< > to access the saveset.  It was transferred in binary mode.  A If you are lazy, then just get a copy of the FIX_SAVESET utility.n  1 (f.ex. from ftp://ftp.hhs.dk/pub/vms/collection/)r   Arne   ------------------------------  # Date: Fri, 26 Jan 2001 15:38:35 GMTc* From: Alan E. Feldman <alan48@my-deja.com>- Subject: Re: ftp'ing savesets from W2K to VMSt) Message-ID: <94s5lq$rf3$1@nnrp1.deja.com>o  9 In article <EJEDILJIPPJOIEFOLDHLIENDCEAA.tom@kednos.com>, $   Tom Linden <tom@kednos.com> wrote:B > I think one needs to change block size.  Can someone give me the command < > to access the saveset.  It was transferred in binary mode. > tia  >u  8 See the OpenVMS FAQ, MGMT20. section. It tells you about@ RESET_BACKUP_SAVESET_ATTRIBUTES.COM which can be obtained on the< Freeware/CD/Webpages and other useful info about this topic.   --F NOTE: If you wish to e-mail me, please do NOT use the deja address. ItD is not a valid address. Instead, use the address below, removing the long wrong part first. Thanks.    Disclaimer: JMHO Alan E. Feldman  &-)( afeldman@gfigroup.ButItSaidItPrinted.com     Sent via Deja.comf http://www.deja.com/   ------------------------------  # Date: Fri, 26 Jan 2001 22:04:31 GMT42 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)- Subject: Re: ftp'ing savesets from W2K to VMS 7 Message-ID: <Pvmc6.141$cu.1115@gazette.loc1.tandem.com>    Tom Linden wrote:uJ : I think one needs to change block size.  Can someone give me the command< : to access the saveset.  It was transferred in binary mode.  H   Please see the OpenVMS FAQ.  Various tools (that determine the savesetJ   blocking factor from within the BACKUP saveset) and that then reset the $   attributes to match are available.  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Fri, 26 Jan 2001 15:21:11 -0800r! From: Tom Linden <tom@kednos.com>l- Subject: RE: ftp'ing savesets from W2K to VMS 9 Message-ID: <EJEDILJIPPJOIEFOLDHLEEPPCEAA.tom@kednos.com>L  I Already have thanks.  A /windows (or whatever name to signify the origin)O0 option might be something to add for the future.   > -----Original Message-----; > From: Hoff Hoffman [mailto:hoffman@xdelta.zko.dec.nospam]O( > Sent: Friday, January 26, 2001 2:05 PM > To: Info-VAX@Mvb.Saic.Coms/ > Subject: Re: ftp'ing savesets from W2K to VMSb >t >u > Tom Linden wrote:hL > : I think one needs to change block size.  Can someone give me the command> > : to access the saveset.  It was transferred in binary mode. >aJ >   Please see the OpenVMS FAQ.  Various tools (that determine the savesetK >   blocking factor from within the BACKUP saveset) and that then reset theb& >   attributes to match are available. >p4 >  --------------------------- pure personal opinion > ---------------------------e1 >    Hoff (Stephen) Hoffman   OpenVMS Engineering  > hoffman#xdelta.zko.dec.com >e   ------------------------------  % Date: Sat, 27 Jan 2001 02:35:54 +0000u) From: Christof Brass <brass@infopuls.com>   Subject: Future of VMS and Alpha, Message-ID: <3A72340A.9367462A@infopuls.com>  
 Strategy A
 ----------  8 Search a big company to support VMS and Alpha. VMS could> eventually be made open source to gain momentum. Could Samsung be this company?  % 1.Would Samsung be interested in VMS?m  ; 2.Is there a chance that Samsung can play a role like Linusa8 Thorvalds in coordinating an open source version of VMS?  ; 3.As far as I know Samsung is one important share holder ofa@ AlphaIncorporation (the other beeing Compaq). AlphaIncorporation< is producing new Alpha versions and designs (I'm not sure if> this is true). And Samsung produced once Alpha chips, chipsets@ and mainboards. I don't know if it is still the case. Is there a. chance that Samsung will concentrate on Alpha?    
 Strategy B
 ----------   Do it yourself.   @ 1.Is there a chance that the current VMS customers pay for a new@ Company whose only task is to let them keep VMS? The new company4 has to maintain VMS and to adjust it to new demands.  / 2.How can all VMS customers be found and asked?v  > 3.Are there people around who would work entirely for this new= company because they believe in their mission? Of course theyl0 will be payed with the money from the customers.  7 4.Is there a legal problem if this new company uses anys: available VMS sources? Is there a chance to defend the new? company against a law suit initiated by Compaq? Does it help toe? defend the new company if it could be proven that otherwise ther9 VMS customers will loose a lot of money because VMS isn't " supported by Compaq appropriately?    
 Next steps
 ----------  > In any case it would be good to get in contact with as many as> possible VMS customers and to ask them about their attitude to8 the situation and the perspectives. Are they inclined to@ migrate? Would they prefer to keep VMS? How much would they like- to pay (in advance, right now, or over time)?t     Basic question --------------  ; Is VMS enough profitable to pay for its own development andy: marketing? I assume that the service divison is profitable8 anyway if it hasn't to transfer money to development and
 marketing?   ------------------------------  % Date: Fri, 26 Jan 2001 09:08:59 -0500s. From: "Jerry Alan Braga" <jabraga@flanagan.ca>F Subject: HELP, Is anyone using VMS as internet SMTP/POP server via UCX4 Message-ID: <Nyfc6.123857$Z2.1539340@nnrp1.uunet.ca>  J I would like to take advantage of this since we now have a direct internet$ link but how to handle the following  K 1) Attachments, UCX can do it but legacy users using VMS mail cannot handles the mime headers; 2) Distribution groups, want to keep them server side based9C     - it works if you create versions of your .DIS files inside ther7 sys$syspecific[UCX_SMTP] but addresses are not filtered K     - ie: if user JOE is both in GROUP1.DIS and GROUP2.DIS vms mail reomvesa* duplicates but going thru SMTP is does notL 3) Would like to use and web interface into mail, so is there an IMAP server( of HTTP mail server that will run on VMS   ------------------------------  # Date: Fri, 26 Jan 2001 14:44:07 GMTd= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)lJ Subject: Re: HELP, Is anyone using VMS as internet SMTP/POP server via UCX0 Message-ID: <009F6B04.FDB0A410@SendSpamHere.ORG>  e In article <Nyfc6.123857$Z2.1539340@nnrp1.uunet.ca>, "Jerry Alan Braga" <jabraga@flanagan.ca> writes: K >I would like to take advantage of this since we now have a direct internetr% >link but how to handle the followingi >rL >1) Attachments, UCX can do it but legacy users using VMS mail cannot handle >the mime headersa  B What!  Your users can't decipher Base64 encoding as they read!  :)    < >2) Distribution groups, want to keep them server side basedD >    - it works if you create versions of your .DIS files inside the8 >sys$syspecific[UCX_SMTP] but addresses are not filtered   Get MX or PMDF.i    L >    - ie: if user JOE is both in GROUP1.DIS and GROUP2.DIS vms mail reomves+ >duplicates but going thru SMTP is does not M >3) Would like to use and web interface into mail, so is there an IMAP server ) >of HTTP mail server that will run on VMS   C If these VMS users have X displays (VMS or otherwise) use Netscape oB for reading MIMEd messages via POP.  I typically read my mail withC VMS mail but, occasionally, somebody will send me a picture that is D of immediate or vital importance that I view (not really, just beingD facetious).  I simply issue a MAIL> MOVE NEWMAIL disk:[user]MAIL.MAID where the disk:[user]MAIL.MAI is another account I've established to use for POP mail with Netscape.'     --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMi             O city, n., 1. a place where trees are cut down and streets are named after them.e   ------------------------------   Date: 26 Jan 2001 14:53:53 GMT' From: david20@alpha2.mdx.ac.uk (D.Webb)aJ Subject: Re: HELP, Is anyone using VMS as internet SMTP/POP server via UCX0 Message-ID: <94s321$lj2$1@aquila.news.mdx.ac.uk>  e In article <Nyfc6.123857$Z2.1539340@nnrp1.uunet.ca>, "Jerry Alan Braga" <jabraga@flanagan.ca> writes: K >I would like to take advantage of this since we now have a direct internetC% >link but how to handle the following  >sL >1) Attachments, UCX can do it but legacy users using VMS mail cannot handle >the mime headersn< >2) Distribution groups, want to keep them server side basedD >    - it works if you create versions of your .DIS files inside the8 >sys$syspecific[UCX_SMTP] but addresses are not filteredL >    - ie: if user JOE is both in GROUP1.DIS and GROUP2.DIS vms mail reomves+ >duplicates but going thru SMTP is does notp  I I wont try and answer the above since it's ages since I used UCX for SMTPhL on its own - we use the commercial PMDF mail product with UCX (which handles) both of these problems and many others). t9 See  http://www.innosoft.com   and http://www.process.comr for more details on PMDF.r    M >3) Would like to use and web interface into mail, so is there an IMAP server ) >of HTTP mail server that will run on VMSi >y   PMDF provides an IMAP server.hK However if you want a web-interface into mail try the freeware YAHMAIL fromn http://wasd.vsm.com.au/wasd/  C This uses the callable mail interfaces to directly access VMS Mail.uL Allowing access to all folders. Handles Mime and attachments. Allows you to H attach files from the system the browser is running on and send them as  attachments.  G This is a CGI which works with just about all the VMS based webservers a (OSU, WASD, APACHE,Purveyor)    
 David Webb VMS and Unix team leader CCSS Middlesex University   ------------------------------  % Date: Fri, 26 Jan 2001 14:47:24 -0500(- From: JF Mezei <jfmezei.spamnot@videotron.ca>(J Subject: Re: HELP, Is anyone using VMS as internet SMTP/POP server via UCX, Message-ID: <3A71D44A.7191D410@videotron.ca>   Jerry Alan Braga wrote:iN > 3) Would like to use and web interface into mail, so is there an IMAP server* > of HTTP mail server that will run on VMS  J Look into PMDF with is an SMTP backbone for VMS that does most of what you want.   K Also, ALL-IN-1 (Office server) has pop, imap, web as well as character cell./ interfaces to email and is highly customizable.i   ------------------------------  # Date: Fri, 26 Jan 2001 16:22:46 GMT1 From: jack9438@my-deja.com& Subject: Internet mail <> OpenVMS Mail) Message-ID: <94s88a$ttc$1@nnrp1.deja.com>d  @ Can someone please describe for me what it might take to give myC OpenVMS users access to internet mail via the OpenVMS mail utility?l  A Basics are: OpenVMS 7.1-1H2 running DECNet Phase IV and UCX v4.2.pE About 40 users total, mostly connected via VT420/520 terminals on LATr terminal servers.t  > I'd appreciate some help, beginning with acheiving an internetH connection (LAN Modem?) through automating the reciept of incoming mail.   Thanks!l     Sent via Deja.coma http://www.deja.com/   ------------------------------  % Date: Fri, 26 Jan 2001 11:58:13 -0500 0 From: David Beatty <David.Beatty.NOSPAM@sas.com>* Subject: Re: Internet mail <> OpenVMS Mail2 Message-ID: <E6txOhisKS1wogWO8HqhJDgDhd+D@4ax.com>  2 Setup SMTP under UCX (SYS$MANAGER:UCX$CONFIG.COM),B then look at UCX help for SET CONFIGURATION SMTP, specifically the /GATEWAY qualifier.   6 Once that's setup, then to mail the addressee would be0 smtp%"inet-email-address"; from DCL, it would be? "smtp%""inet-email-address""".  There is also a way to have VMStB Mail route to SMTP automatically, but I can't remember what it is.   David R. Beattyi  = On Fri, 26 Jan 2001 16:22:46 GMT, jack9438@my-deja.com wrote:   A >Can someone please describe for me what it might take to give myrD >OpenVMS users access to internet mail via the OpenVMS mail utility? >dB >Basics are: OpenVMS 7.1-1H2 running DECNet Phase IV and UCX v4.2.F >About 40 users total, mostly connected via VT420/520 terminals on LAT >terminal servers. >-? >I'd appreciate some help, beginning with acheiving an internet0I >connection (LAN Modem?) through automating the reciept of incoming mail.  >d >Thanks! >M >  >Sent via Deja.com >http://www.deja.com/p   ------------------------------  # Date: Sat, 27 Jan 2001 01:22:41 GMTt6 From: "Andy Bustamante" <A_C_Bustamante@earthlink.net>? Subject: Re: Is Compaq dropping support for vms on vax platform"E Message-ID: <Bppc6.11663$k%3.349330@newsread1.prod.itd.earthlink.net>r  L I turned off our last PDP January 99 as part of my Y2K checklist.  Of courseL I had to put one together again in April and pull some MACRO source down forL the lawyers.  Someone tried to sue on a new geographic "bulleye" patent; but! we'd been doing this for decades.a     -- Andy Bustamantei Remove the ASCII 95s to reply=8 Dave Weatherall <djweath@attglobal.net> wrote in message/ news:DTiotGxQ0bj6-pn2-U1RT12jNt1kE@localhost...rG > On Tue, 23 Jan 2001 12:46:26, David B Sneddon <dbsneddon@bigpond.com>x > wrote: >c@ > > At 11:26 AM 1/23/01 +0000, steven.reece@quintiles.com wrote: > >o > > >Fabio wrote:tB > > >>>>So, if Compaq is dropping VMS on VAX, it is time to have a< > > >VAX on a chip in  a PCI card running under Windows 2000 > > >and CHARON-VAX.....<<<g > >i> > > And just out of curiosity, how many PDPs are still in use? >  > We're down to one! >: > -- > Cheers - Dave.   ------------------------------  % Date: Fri, 26 Jan 2001 15:01:45 -0500-- From: JF Mezei <jfmezei.spamnot@videotron.ca>C- Subject: Re: It's the end for VMS get off now , Message-ID: <3A71D7A6.C040B849@videotron.ca>   John Vottero wrote:e8 > I wonder if this means he intends to sell off OpenVMS?  N Which would be a great thing. However, if Compaq wanted to sell VMS, it shouldL tidy it up a bit and show some growth/momentum to give it value. By ignoringL VMS and keeping it in the dark, Compaq is probably trying to lower the value of VMS.*  L Since Compaq is willing to give away VMS technologies to anyone and everyoneL (Billy Gates), why not just make VMS a GNU-able operating system and be doneL with it once and for all ? Red-Hat would pick it up, market it like hell and kill off Microsoft.*   ------------------------------  % Date: Fri, 26 Jan 2001 15:12:09 +0000*% From: Alan Greig <agreig@my-deja.com> ) Subject: It's the end for VMS get off nowd8 Message-ID: <ll437t095bqp8p20v20j4ervi0lokeg3ru@4ax.com>  C Sorry folks but I have just listened to Capellas speech at analysts-F conference and it could not be clearer that VMS is about to be killed.  F Here's my real time summary of what he said. I've typed it rapidly but here's the gist.   ----+ Well so far it couldn't be more depressing.t  % Typing quickly while still listening.1  E Basically says that Compaq will help Microsoft take over the world. AsC new agreement has been signed to transfer *ALL* of Compaq's cluster+F tech to Microsoft. Plus Compaq will assign engineers to make sure thisD happens. Win2k Data centre will take over the world eventually he isF convinced and Compaq will help msoft attack the unix server market!!!!  @ Will also work with Oracle to make OPS work with clusters on new Windows based systems.  ? Unix on Alphaservers mentioned very briefly. No mention of VMS.o  F He quoted "Only the paranoid survive" - Yep we should get off VMS now!  E Mainframe opsys era is over he claims. Replaced with distributed .NET  and Oracle    D So I don't think Capellas could be clearer. VMS is dead, dead, dead.? Even if it gets mentioned later in speeches by underlings or intB throw-away comments he has effectively announced the death of VMS.1 Only survival might be in few niche applications.   A Capellas says often asked: Why have you got all this other stuff?e (could vms be coming up here?)  D Will always be enterprise data store, massive clusters always up etc (will he mention VMS?)4 Bottom line: You got to be able to spread it across.   We have: 14 out of 15 stock exchanges Himalaya mentioned - [not VMS] 6/7 supercomputer wins
 best clusters  No 1 in SANS
 No 1 in win2k= No 1 in web serversc
 No 1 in linuxR   [Still no VMS!!!]r   Our strategy is:7 Himalaya for non stop and always available data storagee Alpha Unix for supercomputersg) industry standard PCs for everything elseaE If it doesn't fit in with above - we won't do it. [Looks very bad fory	 VMS this],   End of Capellas speech.e   It's over for VMS I say. --   --
 Alan Greig   ------------------------------    Date: 26 Jan 2001 07:23:41 -07001 From: nothome@spammers.are.scum (Malcolm Dunnett)E- Subject: Re: It's the end for VMS get off nowp, Message-ID: <XmwP5I76Uxdo@malvm1.mala.bc.ca>  9 In article <ll437t095bqp8p20v20j4ervi0lokeg3ru@4ax.com>, D*    Alan Greig <agreig@my-deja.com> writes:   > G > Basically says that Compaq will help Microsoft take over the world. AgE > new agreement has been signed to transfer *ALL* of Compaq's clusteriH > tech to Microsoft. Plus Compaq will assign engineers to make sure thisF > happens. Win2k Data centre will take over the world eventually he isH > convinced and Compaq will help msoft attack the unix server market!!!! >   F    Compaq started life as a "me too", making boxes whose only claim to, fame was that they worked just like IBM PCs.  D    It appears their corporate philosophy has never wavered from thatE original principle. They demonstrated time after time that they don'tvD have the vision to know what to do with a product like VMS. It must A be really disheartening to be one of the many folks at Compaq whoc= believe in VMS to see this sort of clueless attitude from thet decision makers.  E    Did any of the analysts ask why, after Compaq gives away the store G to MS, anyone would want to buy Compaq gear to run this brave new world  of MS clusters?e  <    Was Terry there? Didn't he at least mention the "V" word?   ------------------------------  % Date: Fri, 26 Jan 2001 16:51:28 +0000m% From: Alan Greig <agreig@my-deja.com>e- Subject: Re: It's the end for VMS get off nowl8 Message-ID: <dja37tk4dq6rleq26f2knj8nmnjg53vlvq@4ax.com>  A On 26 Jan 2001 07:23:41 -0700, nothome@spammers.are.scum (Malcolme Dunnett) wrote:d    F >   Did any of the analysts ask why, after Compaq gives away the storeH >to MS, anyone would want to buy Compaq gear to run this brave new world >of MS clusters? > = >   Was Terry there? Didn't he at least mention the "V" word?L  D There are still speeches being made by various underlings. None haveC yet mentioned VMS. All have mentioned Himalaya and stock exchanges.iD Questions, I guess, will come at the very end - 5pm Eastern I guess.  D I still expect VMS to get a mention - probably right at the end on a slide.   --
 Alan Greig   ------------------------------  % Date: Fri, 26 Jan 2001 12:58:30 -0500n% From: "John Vottero" <John@mvpsi.com>s- Subject: Re: It's the end for VMS get off nowa/ Message-ID: <t73enbd3gh2pa4@news.supernews.com>   2 "Alan Greig" <agreig@my-deja.com> wrote in message2 news:ll437t095bqp8p20v20j4ervi0lokeg3ru@4ax.com... >    [snip]   >  > Our strategy is:9 > Himalaya for non stop and always available data storage- > Alpha Unix for supercomputersl+ > industry standard PCs for everything else G > If it doesn't fit in with above - we won't do it. [Looks very bad forc > VMS this]- >-  6 I wonder if this means he intends to sell off OpenVMS?   ------------------------------  % Date: Fri, 26 Jan 2001 13:04:30 -0500a% From: "Brian Tillman" <tillman_brian>A- Subject: Re: It's the end for VMS get off nowr" Message-ID: <3a71bb48@news.si.com>  I I'm listening to the Analyst meeting now.  Compaq is expecting to captureuI market in web serving and Internet.  He's said they're after IBM and Sun.eI Capellas also said they're going to be a software company again.  They'renI really wooing Oracle.  Enterprise systems are Tru64.  They're not gettingt@ into a price war on the low end just to gain market share there.  ! Oops.  I just lost my connection.w --A Brian Tillman                   Internet: tillman_brian at si.comgA Smiths Aerospace                          tillman at swdev.si.comi= 3290 Patterson Ave. SE, MS      Addresses modified to preventn< Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"8        This opinion doesn't represent that of my company   ------------------------------  % Date: Fri, 26 Jan 2001 18:46:43 +0000m% From: Alan Greig <a.greig@virgin.net> - Subject: Re: It's the end for VMS get off nowr* Message-ID: <3A71C612.BC5E7F2F@virgin.net>   Brian Tillman wrote:  K > I'm listening to the Analyst meeting now.  Compaq is expecting to capturetK > market in web serving and Internet.  He's said they're after IBM and Sun.pK > Capellas also said they're going to be a software company again.  They'reeK > really wooing Oracle.  Enterprise systems are Tru64.  They're not gettingeB > into a price war on the low end just to gain market share there. >r# > Oops.  I just lost my connection.o >   K I can't seem to get back in at the moment either. Windows Media Player just-L sits there. Didn't hear Capellas say they are going to be a software companyL again but I had to break off for a bit. What does he mean by "again" anyway?  M Still I've now heard three speakers. No mention of VMS or if there was it wasuK so brief I missed it. Lots of mentions for Windows, a number for Himalaya - M he makes out Himalaya is Compaq's entire presence in stock exchanges. Unix ismJ usually only mentioned in terms of Compaq helping Microsoft to "attack the( soft underbelly of Unix" - direct quote.  L Couple of mentions for Alpha supercomputers running Linux. Almost no mention of T64 by name by anyone.r   > --C > Brian Tillman                   Internet: tillman_brian at si.comnC > Smiths Aerospace                          tillman at swdev.si.comn? > 3290 Patterson Ave. SE, MS      Addresses modified to prevent > > Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@": >        This opinion doesn't represent that of my company   --
 Alan Greig   ------------------------------  % Date: Fri, 26 Jan 2001 13:20:49 -0500 % From: "Brian Tillman" <tillman_brian>r- Subject: Re: It's the end for VMS get off nowc$ Message-ID: <3a71bf1b$1@news.si.com>  F Large engines marrying Unix and Windows.  Enhancements of the handheldK devices - iPAQ and wireless, especially in the corporate Email environment.eK The claim is PocketPC leads in this area.  Partnering with telcos is on the C horizon for the 2003-4 time frame to include wireless in just about2' everything.  Won't predict Alpha sales.n   That's the end of the webcast. --A Brian Tillman                   Internet: tillman_brian at si.comOA Smiths Aerospace                          tillman at swdev.si.com = 3290 Patterson Ave. SE, MS      Addresses modified to preventl< Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"8        This opinion doesn't represent that of my company   ------------------------------  % Date: Fri, 26 Jan 2001 13:31:24 -0600b1 From: "Dave Gudewicz" <david.gudewicz@abbott.com>m- Subject: Re: It's the end for VMS get off nowr8 Message-ID: <94sj2k$duh$1@fizban.fizban.pprd.abbott.com>  L I sent email to Capellas this am basically asking if he's thrown his license plate in the scrap heap.  G You'll recall he wore the VMS license plate around his neck at the 1999oI DECUS event in SanDiego, I believe it was.  When pressed on the continuedmD VMS support issue by someone during the Q & A, he left the stage andJ presented his business card to the person that asked the question and said! call me, I support VMS blah blah.M  2 The crowd reacted very positively to that gesture.   Wonder how the crowd feels now.I   Dave...s  2 "Alan Greig" <a.greig@virgin.net> wrote in message$ news:3A71C612.BC5E7F2F@virgin.net... >i >r > Brian Tillman wrote: > E > > I'm listening to the Analyst meeting now.  Compaq is expecting tom capture H > > market in web serving and Internet.  He's said they're after IBM and Sun.D > > Capellas also said they're going to be a software company again. They'renE > > really wooing Oracle.  Enterprise systems are Tru64.  They're notO getting D > > into a price war on the low end just to gain market share there. > > % > > Oops.  I just lost my connection.- > >- > H > I can't seem to get back in at the moment either. Windows Media Player justF > sits there. Didn't hear Capellas say they are going to be a software company F > again but I had to break off for a bit. What does he mean by "again" anyway?. >eK > Still I've now heard three speakers. No mention of VMS or if there was itk wasaB > so brief I missed it. Lots of mentions for Windows, a number for
 Himalaya -L > he makes out Himalaya is Compaq's entire presence in stock exchanges. Unix isL > usually only mentioned in terms of Compaq helping Microsoft to "attack the* > soft underbelly of Unix" - direct quote. >eF > Couple of mentions for Alpha supercomputers running Linux. Almost no mention  > of T64 by name by anyone.s >J > > --E > > Brian Tillman                   Internet: tillman_brian at si.com E > > Smiths Aerospace                          tillman at swdev.si.comeA > > 3290 Patterson Ave. SE, MS      Addresses modified to preventD@ > > Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"< > >        This opinion doesn't represent that of my company >r > -- > Alan Greig >t >s   ------------------------------  % Date: Fri, 26 Jan 2001 14:39:52 -0500e- From: "Peter Weaver" <peter.weaver@stelco.ca>i- Subject: Re: It's the end for VMS get off nowv4 Message-ID: <qokc6.123934$Z2.1541787@nnrp1.uunet.ca>  < "Dave Gudewicz" <david.gudewicz@abbott.com> wrote in message2 news:94sj2k$duh$1@fizban.fizban.pprd.abbott.com... > ... ! > Wonder how the crowd feels now.e  1 Scared, nervous, worried, terrified, petrified...a   --   RULES OF THE AIR   ----------------- <  #19. In the ongoing battle between objects made of aluminum;       going hundreds of miles per hour and the ground going 6       zero miles per hour, the ground has yet to lose.   ------------------------------  % Date: Fri, 26 Jan 2001 14:56:15 -0500r- From: JF Mezei <jfmezei.spamnot@videotron.ca><- Subject: Re: It's the end for VMS get off nowL, Message-ID: <3A71D65D.E8A90FA8@videotron.ca>   Alan Greig wrote:- > E > Sorry folks but I have just listened to Capellas speech at analystsdH > conference and it could not be clearer that VMS is about to be killed.    N Hey Mr Deninger, is that better or worse FUD than what you had accused me of ?   ------------------------------  % Date: Fri, 26 Jan 2001 19:57:32 +0000h% From: Alan Greig <a.greig@virgin.net>c- Subject: Re: It's the end for VMS get off nowz* Message-ID: <3A71D6AC.35BE44E0@virgin.net>   Dave Gudewicz wrote:  N > I sent email to Capellas this am basically asking if he's thrown his license > plate in the scrap heap. >.I > You'll recall he wore the VMS license plate around his neck at the 1999 K > DECUS event in SanDiego, I believe it was.  When pressed on the continuedaF > VMS support issue by someone during the Q & A, he left the stage andL > presented his business card to the person that asked the question and said# > call me, I support VMS blah blah.e >n  G It always struck me that hanging something round your neck has negativetN connotations. The ancient mariner and his albatross. Didn't want to mention it
 until now.   >-4 > The crowd reacted very positively to that gesture. > ! > Wonder how the crowd feels now.e >y	 > Dave.... >>4 > "Alan Greig" <a.greig@virgin.net> wrote in message& > news:3A71C612.BC5E7F2F@virgin.net... > >  > >  > > Brian Tillman wrote: > >$G > > > I'm listening to the Analyst meeting now.  Compaq is expecting tol	 > captureTJ > > > market in web serving and Internet.  He's said they're after IBM and > Sun.F > > > Capellas also said they're going to be a software company again.	 > They'reaG > > > really wooing Oracle.  Enterprise systems are Tru64.  They're note	 > gettingtF > > > into a price war on the low end just to gain market share there. > > >c' > > > Oops.  I just lost my connection.e > > >  > >iJ > > I can't seem to get back in at the moment either. Windows Media Player > justH > > sits there. Didn't hear Capellas say they are going to be a software	 > companyAH > > again but I had to break off for a bit. What does he mean by "again"	 > anyway?r > >-M > > Still I've now heard three speakers. No mention of VMS or if there was it  > wasdD > > so brief I missed it. Lots of mentions for Windows, a number for > Himalaya -N > > he makes out Himalaya is Compaq's entire presence in stock exchanges. Unix > isN > > usually only mentioned in terms of Compaq helping Microsoft to "attack the, > > soft underbelly of Unix" - direct quote. > >nH > > Couple of mentions for Alpha supercomputers running Linux. Almost no	 > mentionr > > of T64 by name by anyone.a > >  > > > --G > > > Brian Tillman                   Internet: tillman_brian at si.comlG > > > Smiths Aerospace                          tillman at swdev.si.comoC > > > 3290 Patterson Ave. SE, MS      Addresses modified to preventcB > > > Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"> > > >        This opinion doesn't represent that of my company > >0 > > -- > > Alan Greig > >t > >    --
 Alan Greig   ------------------------------  % Date: Fri, 26 Jan 2001 20:13:34 +0000t% From: Alan Greig <a.greig@virgin.net>u- Subject: Re: It's the end for VMS get off now * Message-ID: <3A71DA6D.4A147E9E@virgin.net>  8 > I wonder if this means he intends to sell off OpenVMS?  H That would seem to be the only other option that fits in with his speechH other than killing it. It is difficult to see anyone wanting VMS withoutF having Alpha. And really only something like IBM could be interested IK would think. Unless Compaq committed to pay for an IA64 port as part of theCE deal. Yes I know the views on IA64 but better VMS on IA64 than VMS on. nothing.  J Now a sell off to the VMS division itself would have me jumping for joy ifH they got either cast iron guarantees on Alpha future and/or an IA64 port* funded. But that couldn't happen could it?  J I remain confident that Intel will make IA64 mainstream. Maybe take them aK couple of years longer than they expected but if they don't it could be the % end of Intel. That's what's at stake.n --
 Alan Greig   ------------------------------   Date: 26 Jan 2001 20:19:41 GMT* From: bdwheele@indiana.edu (Brian Wheeler)- Subject: Re: It's the end for VMS get off nowk3 Message-ID: <94sm4t$jt5$2@flotsam.uits.indiana.edu>-  * In article <3A71D6AC.35BE44E0@virgin.net>,( 	Alan Greig <a.greig@virgin.net> writes: >  >  > Dave Gudewicz wrote: > O >> I sent email to Capellas this am basically asking if he's thrown his license  >> plate in the scrap heap.d >>J >> You'll recall he wore the VMS license plate around his neck at the 1999L >> DECUS event in SanDiego, I believe it was.  When pressed on the continuedG >> VMS support issue by someone during the Q & A, he left the stage andeM >> presented his business card to the person that asked the question and saidt$ >> call me, I support VMS blah blah. >> > I > It always struck me that hanging something round your neck has negativeeP > connotations. The ancient mariner and his albatross. Didn't want to mention it > until now. >     4 Not to mention prisoners with their number cards....   Brianw   >>5 >> The crowd reacted very positively to that gesture.  >>" >> Wonder how the crowd feels now. >>
 >> Dave... >>5 >> "Alan Greig" <a.greig@virgin.net> wrote in messager' >> news:3A71C612.BC5E7F2F@virgin.net...e >> > >> > >> > Brian Tillman wrote:s >> >H >> > > I'm listening to the Analyst meeting now.  Compaq is expecting to
 >> captureK >> > > market in web serving and Internet.  He's said they're after IBM and, >> Sun.aG >> > > Capellas also said they're going to be a software company again.o
 >> They'reH >> > > really wooing Oracle.  Enterprise systems are Tru64.  They're not
 >> gettingG >> > > into a price war on the low end just to gain market share there.o >> > >( >> > > Oops.  I just lost my connection. >> > > >> >K >> > I can't seem to get back in at the moment either. Windows Media Player  >> justnI >> > sits there. Didn't hear Capellas say they are going to be a softwaree
 >> companyI >> > again but I had to break off for a bit. What does he mean by "again"i
 >> anyway? >> >N >> > Still I've now heard three speakers. No mention of VMS or if there was it >> wasE >> > so brief I missed it. Lots of mentions for Windows, a number fors
 >> Himalaya - O >> > he makes out Himalaya is Compaq's entire presence in stock exchanges. Unixw >> is O >> > usually only mentioned in terms of Compaq helping Microsoft to "attack the - >> > soft underbelly of Unix" - direct quote.- >> >I >> > Couple of mentions for Alpha supercomputers running Linux. Almost noe
 >> mention >> > of T64 by name by anyone. >> >	 >> > > --eH >> > > Brian Tillman                   Internet: tillman_brian at si.comH >> > > Smiths Aerospace                          tillman at swdev.si.comD >> > > 3290 Patterson Ave. SE, MS      Addresses modified to preventC >> > > Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@" ? >> > >        This opinion doesn't represent that of my company  >> > >> > --s >> > Alan Greigs >> > >> > >  > -- > Alan Greig >  >    ------------------------------   Date: 26 Jan 2001 20:18:36 GMT* From: bdwheele@indiana.edu (Brian Wheeler)- Subject: Re: It's the end for VMS get off nows3 Message-ID: <94sm2s$jt5$1@flotsam.uits.indiana.edu>l  , In article <3A71D7A6.C040B849@videotron.ca>,0 	JF Mezei <jfmezei.spamnot@videotron.ca> writes: > John Vottero wrote:e9 >> I wonder if this means he intends to sell off OpenVMS?  > P > Which would be a great thing. However, if Compaq wanted to sell VMS, it shouldN > tidy it up a bit and show some growth/momentum to give it value. By ignoringN > VMS and keeping it in the dark, Compaq is probably trying to lower the value	 > of VMS.  > N > Since Compaq is willing to give away VMS technologies to anyone and everyoneN > (Billy Gates), why not just make VMS a GNU-able operating system and be doneN > with it once and for all ? Red-Hat would pick it up, market it like hell and > kill off Microsoft.g  J There are certain charms to this, especially _after_ M$ has paid the money for the cluster information :)  J It also might end the constant whining for an x86 port...at least it would0 allow a quick "do it your damned self" answer :)  M I've often wondered why they don't do this.  VMS source is already available,CJ so its not a super-secret.  They'd still get their money for support, etc.  N Of course, I suppose it all depends on the weather:  when hell freezes over :(  
 Brian Wheeleru bdwheele@indiana.edu   ------------------------------    Date: 26 Jan 2001 15:49:00 -0500* From: young_r@eisner.decus.org (Rob Young)- Subject: Re: It's the end for VMS get off nowt+ Message-ID: <w0YLK2wZf78T@eisner.decus.org>n  R In article <3A71DA6D.4A147E9E@virgin.net>, Alan Greig <a.greig@virgin.net> writes: >  >  > 9 >> I wonder if this means he intends to sell off OpenVMS?: > J > That would seem to be the only other option that fits in with his speechJ > other than killing it. It is difficult to see anyone wanting VMS withoutH > having Alpha. And really only something like IBM could be interested IM > would think. Unless Compaq committed to pay for an IA64 port as part of thesG > deal. Yes I know the views on IA64 but better VMS on IA64 than VMS ons
 > nothing. > L > Now a sell off to the VMS division itself would have me jumping for joy ifJ > they got either cast iron guarantees on Alpha future and/or an IA64 port, > funded. But that couldn't happen could it? > L > I remain confident that Intel will make IA64 mainstream. Maybe take them aM > couple of years longer than they expected but if they don't it could be then' > end of Intel. That's what's at stake.k  B 	They would have to get a tremendous premium in a sell off though.A 	After all, doesn't VMS contribute nice *profits*?  Nasty little  
 	catch-22.   				Rob    ------------------------------  % Date: Fri, 26 Jan 2001 16:03:21 -0500w- From: JF Mezei <jfmezei.spamnot@videotron.ca>r- Subject: Re: It's the end for VMS get off nowr, Message-ID: <3A71E612.EB1E1092@videotron.ca>   Brian Wheeler wrote:P > Of course, I suppose it all depends on the weather:  when hell freezes over :(  0 Have you checked the weather in Siberia lately ?  R (reminiscent of the "Have you checked Discovery's orbit lately" in the movie 2010)    N Since hell has in fact frozen over in Siberia (with temperatures near -53c), IO think that there is no longer any restrictions on what could be done with VMS.    K I for one would much prefer VMS being set free, even if this means a lot ofiN uncertaintly, compared to VMS being tied to a large post on top of a wood pile% with a fire smoldering underneath it.n  L The biggest difficulty would be to find a way to keep the good VMS engineersI during the period where VMS would have no revenus and the period where itEN starts to get them. (during that time, VMS would have to be essential "stable" at 7.3)t   ------------------------------  % Date: Fri, 26 Jan 2001 16:11:38 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca>h- Subject: Re: It's the end for VMS get off nown, Message-ID: <3A71E803.696A6269@videotron.ca>   Rob Young wrote:K >         They would have to get a tremendous premium in a sell off though.tI >         After all, doesn't VMS contribute nice *profits*?  Nasty littleu >         catch-22.t  N What was the value of Titanic after they realised it would sink within 2 hoursN ? I was still generating electricity, its dining rooms were still functioning,P and generally, most of the ship was unaware of the impending doom for some time.  L For Compaq, it makes good business sense to sell VMS. Compaq maintains AlphaN sales (at least until other manufacturers start to make VMS compatible alphas)N and Compaq maintains the maintenance revenus for VMS sites (new and exsiting).  K And Compaq would probably still get some revenus from selling stuff such asd- Compilers and TCPIP stack to the VMS company.   F But Compaq would not have to be bothered with selling this systems, no& marketing, no sales training etc etc.   I The new company can do like RedHat and sell from a central point, and getB> commissions from al the support contracts they send to Compaq.   ------------------------------  # Date: Sat, 27 Jan 2001 00:17:02 GMTr= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) - Subject: Re: It's the end for VMS get off nowE0 Message-ID: <009F6B55.06CC8572@SendSpamHere.ORG>  l In article <94sj2k$duh$1@fizban.fizban.pprd.abbott.com>, "Dave Gudewicz" <david.gudewicz@abbott.com> writes:M >I sent email to Capellas this am basically asking if he's thrown his license. >plate in the scrap heap.  > H >You'll recall he wore the VMS license plate around his neck at the 1999J >DECUS event in SanDiego, I believe it was.  When pressed on the continuedE >VMS support issue by someone during the Q & A, he left the stage and"K >presented his business card to the person that asked the question and said-" >call me, I support VMS blah blah. >,3 >The crowd reacted very positively to that gesture.d >d  >Wonder how the crowd feels now. >I >Dave...  H As just one of many in attendance at that event, I can only say that I'mH not going to go to sleep tonight with a warm and fuzzy for the future of
 VMS.  :(    4 How many times did the boy in the fable "cry wolf"??   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COM.            dO city, n., 1. a place where trees are cut down and streets are named after them.a   ------------------------------  % Date: Fri, 26 Jan 2001 19:26:59 -0700n1 From: Vance Haemmerle <vance@toyvax.Tucson.AZ.US>r- Subject: Re: It's the end for VMS get off now 3 Message-ID: <3A71CF83.4A80898C@toyvax.Tucson.AZ.US>    Peter Weaver wrote:n > > > "Dave Gudewicz" <david.gudewicz@abbott.com> wrote in message4 > news:94sj2k$duh$1@fizban.fizban.pprd.abbott.com... > > ...e# > > Wonder how the crowd feels now.a > 3 > Scared, nervous, worried, terrified, petrified...d  B   Selling their Compaq stock?  Capellas is as clueless as Ekhard. F I'm glad I sold my 1100 shares of CPQ last May for $26/share.  Nothing' has happened that has changed my mind. e   --B Vance Haemmerle               Internet   vance@toyvax.Tucson.AZ.USK Tucson, AZ                    Web        http://toyvax.Tucson.AZ.US/~vance/    ------------------------------  % Date: Fri, 26 Jan 2001 22:38:16 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca> - Subject: Re: It's the end for VMS get off nowo, Message-ID: <3A7242A7.60084403@videotron.ca>  & "Brian Schenkenberger, VAXman-" wrote:6 > How many times did the boy in the fable "cry wolf"??  L But if you have a bad flue, and your mother tucks you to bed every night andK tells you "sleep tight, the flue will go away soon" and the flue never goess* away, you will stop beleiving your mother.   Cuts both ways.   M What Compaq did this week ranks high up there with the Palmer era advertisingo& to dump you legacy VMS and go to Unix.   ------------------------------  # Date: Fri, 26 Jan 2001 08:37:46 GMTa From: sabolich@my-deja.com- Subject: Re: Linux worm and RedHat 7.0 brokenA) Message-ID: <94rd0q$7jb$1@nnrp1.deja.com>t   snip  H > Well, file specs in VMS are richer than in unix, and I realize that itH is fairly common for unix programmers to do it by hand.  A full VMS file spec looks something like this: I >   node"username password"::device:[dir1.dir2.dir3]filename.type;version-? > (Any of this can be omitted to take the defaults, of course.)WG > This is more than trivial syntax changes from unix.  There's a lot offC semantics.  There are 3 kinds of wildcards, which are a little more @ complex than in unix.  Several parts can be logical names, whichC sometimes act like wildcards.  The standard parser allows up to twooB levels of component-by-component defaulting.  And there are littleH backward compatibility things like allowing <> instead of [], and period5 instead of semicolon.  And I'm omitting many details.e >oG > Virtually everyone who tries to do this parsing by hand gets at leastoG some of it wrong, and ends up with an application that doesn't work foraG some perfectly valid file specs.  If you let the system do the work, ituF behaves "right" for anything the user throws at it.  You will say thisF is a terribly complex file naming system, and it is if you use it all.< But it was all built up over the years to provide real-worldE functionality.  As features were added, the tools got smarter withoutn> breaking compatibility, and existing applications kept working> automagically.  That's good design.  See other comments below.  F Ok, I understand.  We (the VMS architects) reserve the right to changeD the file naming system - so don't do hand parsing.  Use the suppliedD parsing routines because they will compensate for any changes in the! file naming system.  Interesting.c   So Unix way would be:p  5 mount node"username:password"::device:[dir1] /somedir   # then simply use /somedir/dir2......v   snip  G > VMS has improved directory caching lately to help with such problems.sC  More improvements are in the pipeline, IIRC.  In the meantime, VMSiD folks usually distribute their files among more directories. LogicalF names make this transparent in many cases, and the intelligent parsingG helps too.  Soemone here can probably suggest improvements, if you make # the problem a little more specific.t   Thanks.i   >.G > For existing huge directories, suggest to your guy that he delete therD files in reverse order, or use one of the freeware tools that deleteI many files without promising moment-by-moment integrity of the directory.  > 0 Thanks again.  Basically async metadata updates?   snip  G > > "looking out the window at coeds". Please, this is not a flame war.t$ > > Let's keep it from becoming one. >tF > Ok.  I just have trouble coming up with a mental image of those unixH guys designing.  I was only guessing about possible alternatives.   I'll4 put in a few extra smileys just in case. :-) :-) :-)  8 :-) Do you have a mental image of VMS guys designing :-)  I > > You asserted that you need to change kernel source code and recompile L > > the kernel for "simple" changes in OS configuration.  Do you still stand > > by this? >eD > I'd don't.  The guys down the hall do, frequently.  Maybe they are3 doing it wrong, but they claim to be linux experts.e >eE Many people do because they can.  For example usually Linux kernel iskE shipped with FPU emulation compiled in for 386 or 486sx.  Some peopleeH with 486DX or better want to reclaim a few K of ram and recompile.  SomeG are like overclockers who try to squeeze out the last few Hz and try tos squeeze out the last few K.i  B Anyways drivers can be either compiled into the kernel or modules.E The most common method is modules.  They can be loaded or unloaded atr@ will.  For example 'rmmod maestro' will unload the driver for myF soundcard (provided that the sound card is not in use at that moment).C 'insmod maestro' will load it again.  Same with nework for example.   % ifdown eth0		// bring down ethernet 0  rmmod rtl8139		// unload driverd$ insmod newrtl8139	// load new driver% ifup eth0		// bring up the ethernet 0p    So you rarely have to reboot it.  = parameter setting is done through /proc filesystem - example:   , echo 7 >/proc/sys/net/ipv4/some_tcp_paramter   snip  K > > Same for Linux -- same binary will run on kernel 2.0.x, 2.2.x, 2.4.x on< > > 386, 486, 586 686. >cF > I believe linux is ahead of most unixes here, which is not uncommon.) But what about linux 1.whatever binaries?  >e; Not sure, but I would be not surprised if the answer is no.h   Fran     Sent via Deja.com  http://www.deja.com/   ------------------------------   Date: 26 Jan 2001 19:25:22 GMT1 From: bill@triangle.cs.uofs.edu (Bill Gunshannon)n- Subject: Re: Linux worm and RedHat 7.0 brokend+ Message-ID: <94siv2$6uu$1@info.cs.uofs.edu>l  L In article <rdeininger-2401012205040001@user-2ive7g7.dialup.mindspring.com>,5  rdeininger@mindspring.com (Robert Deininger) writes:  |> n< |> And "theoretically infinte" strings are a good thing? :-)  B Arbitrary builtin limits are definitely a bad thing.  But then, weB all know that we will never need really big strings.  Just like weD we never needed big disks, a lot of memory or even a comouter in our	 own home.    |> nN |> Hmm.  Neither is perfect, so we should pretend they are of equal quality?  J |> I do think there are meaningful comparisons to be made.  I listed a fewH |> off the top of my head.  Maybe you prefer other points of comparison.  G A different view is not necessarily a bad design nor of poorer quality.n Just different.   G |> > Matter of opinion, I guess.  I find it infinitely easier to modifynH |> > a Unix kernel than a VMS one.  As a matter of fact, I have recentlyG |> > done some work with modified Ultrix-11 kernels and I am doing thiseH |> > with no documentation at all.  I can't do the same on VMS even with |> > a shelf full of manuals.n |> dN |> New glasses, maybe?  :-)  Maybe you aren't really trying to understand VMS.  G No, I find much of it written and layed out n a confusing manner. Maybe F someone forgot that the person who needs the docs the most is the one I who isn't necessarily familiar with the in house jargon.  Especially whennI it differs from how others in the industry say things.  But then, VMS has & nothing on Windows in this department.   |> rC |> It probably is easier to modify a unix kernel.  But to modify itgD |> correctly?  I dunno.  We should also consider why it NEEDS to be ) |> modified so darned much.  Good design?i  @ Who said it had to be modified a lot??  I usually do it once per= machine.  If all the machines I had were exactly the same in l? configuration I would need to do it once.  VMS is no different.e@ Or don't you agree that the amount of memory and the size of the@ disks can require different settings for things like pagefiles??    L |> > Unix versions as well.  Add into that the cross-Unix compatability thatK |> > has been built into most of them today.  FreeBSD running SCO binaries, $ |> > Linux running Solaris binaries. |> ># |> Does Solaris run Linux binaries?>  F Take that up with Andrew.  There is no reason why it can't except thatH Sun doesn't want it to.  All the necessary libraries for Linux are free.   |> gR |> My most recnet experience was with HPUX, and upgrades required rebuilding code.  A What's your point??  So HPUX, a proprietary version of Unix who'shB greatest driving factor is to make it as hard as possible for it's@ current customers to migrate away, does things badly.  Just likeB SunOS and Solaris, I dumped HPUX and IRIX and would never considerC using them again.  But not because of Unix design flaws.  Quite therA opposite.  These proprietary Unix clones tend to deliberately do yA thigs differently in order to set themselves apart in the market. @ In doing so, they frequently abandon good ideas in favor of moreB inferior, but more proprietary ones.  I know that there are peopleC who will disagree with me, but Sun's biggest mistake was to abandon @ SunOS, which was based on BSD Unix, a continuing development, inC favor of SYSV based Solaris when it was obvious that AT&T had givend/ up on any further development of the SYSV line.    |> > |>    : |> > |> Is there a unix hiding somewhere that can do this? |> >  9 |> > Just about any of them that I have ever worked with.  |> -Q |> Hmm.  What's the oldest working binary on your favorite system?  Early 1980's?a  C I'll have to concede that one to you.  Having changed from Sparc to G x86 I have no old binaries for the architecture I am currently running.7   |> B |>  J |> > Maybe, but I spend a lot of time on Unix groups as I have to maintainE |> > more Unix Servers than VMS servers and I can assure you the Unix I |> > advocates don't devote .01% of the time slaming VMS or any none Unixi5 |> > OS that the VMS people here spend slamming Unix.t |>  G |> Well, I've NEVER gone to a unix group to bash unix.  Why do you unixrH |> folks come over here and say thing like "VMS's design isn't clean".   |> (Not you, I realize.)  D Well, thank you for that.  It's true, I don't come here to bash VMS.A I have to maintain VMS servers just like I maintain Unix servers.R@ Both have their strengths and their weaknesses.  But the biggest? difference is in worldview.  Different does not imply better or  worse, just different.     |> eD |> My primary aim is to counter some of the NONSENSE about VMS that E |> I found in this thread.  I tend to assume ignorant nonsense can be J |> cured by removing the ignorance.  Malicious nonsense is another matter.  A That's all well and good, but my intent in this particular threadrA was to counter some of the NONSENSE about Unix that also seems tor@ pop up whenever this kind of thread rears it's ugly head.  It's ? like the arguments over which programming language is the best.0= The whole world isn't a nail, so we need more tools than just:	 hammers. n   bill   -- fJ Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   h   ------------------------------  % Date: Sat, 27 Jan 2001 04:23:33 +0000I) From: Christof Brass <brass@infopuls.com>t- Subject: Re: Linux worm and RedHat 7.0 broken , Message-ID: <3A724D45.92B23127@infopuls.com>   Bill Gunshannon wrote: > N > In article <rdeininger-2401012205040001@user-2ive7g7.dialup.mindspring.com>,7 >  rdeininger@mindspring.com (Robert Deininger) writes:l  N > |> Hmm.  Neither is perfect, so we should pretend they are of equal quality?L > |> I do think there are meaningful comparisons to be made.  I listed a fewJ > |> off the top of my head.  Maybe you prefer other points of comparison. > I > A different view is not necessarily a bad design nor of poorer quality.u > Just different.c   We need to know your criteria!  I > No, I find much of it written and layed out n a confusing manner. MaybelG > someone forgot that the person who needs the docs the most is the oneiK > who isn't necessarily familiar with the in house jargon.  Especially when,K > it differs from how others in the industry say things.  But then, VMS has ( > nothing on Windows in this department.  7 If things *are* different they can't be called with the @ widespread terms. And if you do this kind of system's management( you better got familiar with the jargon.  B > Who said it had to be modified a lot??  I usually do it once per> > machine.  If all the machines I had were exactly the same inA > configuration I would need to do it once.  VMS is no different. B > Or don't you agree that the amount of memory and the size of theB > disks can require different settings for things like pagefiles??  > But amount of memory and size of page- and swapfiles shouldn't@ be compiled in anyway. And what about adjustments to the average@ usage of the system? Is there anything like the VMS optimisation@ cycle of running the system and then running AUTOGEN? Is there a@ usage profile collecting mechanism? Will you re-build the kernel after each change of usage?d  % > |> Does Solaris run Linux binaries?  > H > Take that up with Andrew.  There is no reason why it can't except thatJ > Sun doesn't want it to.  All the necessary libraries for Linux are free.  = Come on, this wasn't the question! Is Solaris able to execute @ Linux binaries? To be able to do that the loader must understand all the Linux binary formats.   C > What's your point??  So HPUX, a proprietary version of Unix who'suD > greatest driving factor is to make it as hard as possible for it'sB > current customers to migrate away, does things badly.  Just likeD > SunOS and Solaris, I dumped HPUX and IRIX and would never considerE > using them again.  But not because of Unix design flaws.  Quite theqB > opposite.  These proprietary Unix clones tend to deliberately doD > things differently in order to set themselves apart in the market.B > In doing so, they frequently abandon good ideas in favor of moreD > inferior, but more proprietary ones.  I know that there are peopleE > who will disagree with me, but Sun's biggest mistake was to abandonmB > SunOS, which was based on BSD Unix, a continuing development, inE > favor of SYSV based Solaris when it was obvious that AT&T had given 1 > up on any further development of the SYSV line.n  = Good point. But I don't think this does any good in improvingm> the credibility of people responsible for and developing UNIX.9 We shouldn't confuse the situation at SUN and HP with thee9 current stupid behaviour of Compaq wrt to VMS. Compaq hasf> alternatives whereas SUN and HP only had their respective OSs.@ And I heard that HP/UX even offers unique realtime qualities. So@ there is at least a reason for doing proprietary extensions. But= what I don't understand and count as point against the way of ? UNIX people's thinking is that the developers at the respectivee? companies deliberately made their OSs worse. How can we explain  that?   S > |> Hmm.  What's the oldest working binary on your favorite system?  Early 1980's?  > E > I'll have to concede that one to you.  Having changed from Sparc touI > x86 I have no old binaries for the architecture I am currently running.d  > We shouldn't stretch this point as it is not clear that beeing? able to execute old binaries is that a good thing. Basically it ? prevents the OS to evolve. If you look at the Windoze API afterw< the evolution from 16 to 32 bit you see what the approach of< backward compatibility and introduction of new features will create: a complete mess.  ? The problem with the UNIX kernel re-build is that if you change.; some data structures which are used by programs as I had tof@ because the structure to hold a host name was too short you will= even make a current binary which is linked against the kerneln? unexecutable. In fact it will execute as with all UNIX variants-< I know of there is no mechanism to detect that the size of a> running kernel structure is not the size of the structure when$ the program was compiled and linked.  L > |> > Maybe, but I spend a lot of time on Unix groups as I have to maintainG > |> > more Unix Servers than VMS servers and I can assure you the UnixoK > |> > advocates don't devote .01% of the time slaming VMS or any none Unixr7 > |> > OS that the VMS people here spend slamming Unix.t > |>I > |> Well, I've NEVER gone to a unix group to bash unix.  Why do you unixtH > |> folks come over here and say thing like "VMS's design isn't clean". > |> (Not you, I realize.) > F > Well, thank you for that.  It's true, I don't come here to bash VMS.C > I have to maintain VMS servers just like I maintain Unix servers. B > Both have their strengths and their weaknesses.  But the biggestA > difference is in worldview.  Different does not imply better orb > worse, just different.  = Good answers but there is still one open point. Do VMS peopled? really slam UNIX more than UNIX people slam VMS or other OSs? Io@ doubt that. If you read the bullshit which is supplied with each= Linux distribution called "VMS to Linux HOWTO" then you mighti? re-think. And it's not only important how many people spend howV= many of their words for slaming (which percentage) it is alsor? important how strong they slam. If you read the bullshit in the:8 "VMS to Linux HOWTO" you don't need much more slaming to@ outweigh all slaming what has and will be done against UNIX from8 VMS people. Moreover the normal UNIX slamer doesn't know; anything about VMS. The normal VMS slamer knows a lot aboutp? UNIX. Of course you are not a normal UNIX against VMS slamer ifc& you are a slamer at all which I doubt.  E > |> My primary aim is to counter some of the NONSENSE about VMS thatsG > |> I found in this thread.  I tend to assume ignorant nonsense can betL > |> cured by removing the ignorance.  Malicious nonsense is another matter. > C > That's all well and good, but my intent in this particular threadlC > was to counter some of the NONSENSE about Unix that also seems toyA > pop up whenever this kind of thread rears it's ugly head.  It'syA > like the arguments over which programming language is the best.c? > The whole world isn't a nail, so we need more tools than justw
 > hammers.  ; Good point but a lot of people use this to escape from realn; analysis. Unfortunately there *is* a real difference beyondt; sympathy or acquaintance between different OSs and PLs. And ; techies should be able to clearly seperate the points where 9 taste comes in and the points where engineering comes in.i   > bill >  > --L > Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesF > bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. > University of Scranton   |@ > Scranton, Pennsylvania   |         #include <std.disclaimer.h>   ------------------------------   Date: 26 Jan 2001 08:51 CSTs' From: carl@gerg.tamu.edu (Carl Perkins) 5 Subject: Re: Memo:  Re:  Memo:  Help! Zip/Unzip errora- Message-ID: <26JAN200108511534@gerg.tamu.edu>   0 Paul BEAUDOIN <paul.beaudoin@hsbc.com> writes... }>:Gents }>A }> Somewhat presumptuous of your potential audience genetics, eh?ek }.........................................................................................................^i } 	 } gender?sF }In any case , yes - modern courtesy has dictated what may not be said }but has yet to supply }a suitable replacement.@ }Various alternative salutations are inappropriate, insulting or
 }unwieldy: }     Sir or Madam }     Ladies and Gentlemen$ }Or how about something a little PC? }     Person }     People }Or even really PC...i8 }     Human (Though even this is a bit presumptuous now) }Or industry specific:, }     Dear Guru/Technosquid/propellorhead...D }I come from a school that demands a salutation as to not use one (aD }possible solution) is akin to barging into a conference unannouncedB }and commandeering the discussion - not generally considered to be
 }sociable.C }So - as in many things - I am a bit stuck on this one. SuggestionsjC }welcome (though I acknowledge this is a technical forum - not Misss
 }Manners.) }  }Pauls   How about any of these:t   Hello, Hi,u Good morning/afternoon/evening,c
 Greetings, Salutations, Salve, To whom it may concern,u What ho!% Come about and prepare to be boarded.d   --- Carl   ------------------------------  # Date: Fri, 26 Jan 2001 16:38:38 GMTO" From: fooguy <jweisen@my-deja.com>7 Subject: Re: Midrange I/O with VMS WAS: DS20 - Slow I/Od) Message-ID: <94s96c$utl$1@nnrp1.deja.com>o  , In article <94q6qn$9s3@gap.cco.caltech.edu>,&   mathog@seqaxp.bio.caltech.edu wrote:H > In article <94pqim$r1u$1@nnrp1.deja.com>, fooguy <jweisen@my-deja.com> writes:sE > >The SSD is a StorageWorks module DS-EZ832 mounted in a BA356 shelfnD > >with a -FA personality module.  The shelf is connected to a KZPCCG > >(Symbios53c895) controller which is capable of 80MB/s LVD operation.a  > >That i/o subsytem looks like: > > F > >KZPCC (80MB/s LVD) --> BA356-FA (20MB/s Ultra) --> DS-EZ832 (40MB/s > >UW) >iH > How big is the SSD?  How much memory will a DS20 hold (4 Gb isn't it?)D > Could you live with putting these tables on a RAMDisk in the DS20s > physical memory?  F The SSD is 3GB. The DS20 will hold 4GB. No, I am not putting an entireH Oracle DB in memory, what if the OS hiccups? Boom. I'm hesitant to ALTER TABLE foo CACHE;  F > I just posted some timing results where a 8326144 byte file was read/ > from and rewritten to RAMDISK on a DS10 usinga >e >   $ copy fromfile tofile >4E > in .008 seconds.  That's 8326144/.008 = 1,040,768,000 bytes/second.nH > You're NEVER going to get performance like that out of anything on theE > other end of a disk controller!  You said the table was 5Gb, if theaG > tables in question are <3.5 Gb or so you'd probably be best served bys > using the RAMdisk.  ; The entire DB is less than 5GB (it's actually less than 3).-  E > Also, the DS20 has twice the memory bandwidth of the DS10, and mosttC > of your accesses will be pure reads, so you could conceivably gete3 > nearly 4Gb/second read speeds off of the RAMdisk.o  2 That's interesting, something I hadn't thought of.   > Just a thought...p >gG > >Let's play "spot the bottleneck", shall we?  In that config, the SSDAD > >is delivering ~13MB/s copying from dir --> dir within the device.> > >Notethat there are no physical moving parts (head movement,B > >rotational latency) to deal with.  Given the BA356-FA, that's a > >pretty good rate;C > >given the $10,000/GB price of the EZ832, I want all 40MB/s it is G > >capable of.  This i/o subsystem is destined to accomodate the Oracle>9 > >indexes and the tables showing the greatest hot spots.  >t> > I think DS20 memory from third parties costs less than that. >  > >m& > >The other i/o subsystem looks like: > >l= > >KZPAC-CA (20MB/s/channel) --> internal split shelf --> 4 x> > >DS-RZ1DD-VW > > F > >At the fastest (RAID 10 across two channels) we are seeing ~1.5MB/sC > >copying dir --> dir within the array and ~6.5MB/s in a BACKUP to F > >NLA0: Given that a two channel array should have a peak performanceG > >of 40MB/s (to NLA0:), this is pathetic.  For comparison, a Sun Netra C > >T1-105 with a single disk delivers ~39MB/s TARring to /dev/null,l1 > >which is about all a single drive can sustain.i >rE > Careful, are you sure that actually read the data from the disk, orhF > did it  come out of file cache.  You have to really mess around withA > RMS parameters to optimize BACKUP performance, how high did youoE > set /BUFFER and /BLOCK?  I also vaguely recall somebody saying thatnH > the  NLA0: device isn't the speediest thing in the world, although youF > really don't have much choice there.  (Be nice if there was some wayH > to set up a RAMdisk so that it was write only, that is, it looked like8 > a huge disk, but what you wrote to it just vaporized.) >nF > >The entire system only needs to support <20 users and less than 5GBF > >of data in the production database; cost-justifying a larger system > >will be quite difficult.o >eA > Buy memory, lots of memory, and you'll have a very fast system.t >g
 > Regards, >s > David Mathog > mathog@seqaxp.bio.caltech.edut@ > Manager, sequence analysis facility, biology division, Caltech   --- *********************************************L( "All I every wanted from life was to see, Larry Wall give Bill Gates a Perl Necklace."     Sent via Deja.comi http://www.deja.com/   ------------------------------  % Date: Fri, 26 Jan 2001 07:16:20 -0500n" From: "Hal Kuff" <kuff@tessco.com>$ Subject: Re: Moving Datatrieve filesO Message-ID: <0D69D7C5D16816DA.2329BF5BD991F8C1.23E4E001B2003701@lp.airnews.net>s  I see the extract command......then just @extract_file from the new version     < "Phillip D. Williams" <edhouse00@qwest.net> wrote in message, news:aX6c6.2130$8q.602354@news.uswest.net... > Hello / > I have a system with VMS 5.5-2 and Datatrieveb7 > 4.3. I have finally decide to do-waway with the 5.5-2e; > system and just install 7.2. Question is since I will notf4 > be doing a upgrade, how do I get Datatrieve 7.1 to= > read the old 4.3 files?? I really dont feel like recreating . > all the Datatrieve domains,definitions, etc. > Thanks > p.s.6 > Has anyone ever went from Excel/Access to Datatrieve > or the other way around???	 > phillips >o >S   ------------------------------  % Date: Fri, 26 Jan 2001 07:43:03 -0500a/ From: "Joe H. Gallagher" <dtrwiz@ix.netcom.com> $ Subject: Re: Moving Datatrieve files- Message-ID: <3A7170D6.E0BDBBBC@ix.netcom.com>e   Phillip D. Williams wrote: >  > Hellou/ > I have a system with VMS 5.5-2 and Datatrieve-7 > 4.3. I have finally decide to do-waway with the 5.5-2i; > system and just install 7.2. Question is since I will noto4 > be doing a upgrade, how do I get Datatrieve 7.1 to= > read the old 4.3 files?? I really dont feel like recreatingB. > all the Datatrieve domains,definitions, etc. > Thanks > p.s.6 > Has anyone ever went from Excel/Access to Datatrieve > or the other way around???	 > phillipv  9 There are two situations.  Do you still have the softwareU4 distribution for the CDD on the 5.5-2 system or not?  C If you do have the software, then save the file sys$system:CDD.DIC.h  = Upgrade the system.  Reinstall the old CDD on the 7.2 system; A Restore the CDD.DIC file.  Install the 7.2 DTR using the options  3 which connect it to the old CDD.  And you are done.   9 If you do not have the CDD software or you do not want toe: use it on the 7.2 system, then for each node on the 5.5-2 ; system where there are dictionary objects you want to move,m: create a file for each dictioary node on the old system by    	DTR> (set dictionary to a node)" 	DTR> extract all on node_name.def   Backup up all these files.  9 Install 7.2, and DTR V7.2.  Configure the dictionary filen< tree for the CDD-less DTR.  Restore the set of node_name.def files, then   * 	DTR> (set dictionary to appropriate node) 	DTR> @node_name.def  : where "node_name" corresponds to the each node of the old  5.5-2 CDD system.t  
 Good luck.   Joe H. Gallagher, Ph. D.$ Former SIG Chair & Newsletter Editor) DATATRIEVE/4GL SIG of DECUS\\\\\Encompass  dtrwiz at ix dot netcom dot comr" See "The DATATRIEVE Programmer" at2 http://www.geocities.com/SiliconValley/Pines/8958/   ------------------------------  % Date: Fri, 26 Jan 2001 20:31:17 +0100e2 From: martin@radiogaga.harz.de (Martin Vorlaender)! Subject: Re: MPACK/MUNPACK imagesi; Message-ID: <3a71d085.524144494f47414741@radiogaga.harz.de>   8 Miller, Daniel (Daniel.Miller@nightfreight.co.uk) wrote:L > I had a few problems with the images. I tried downloading MPACK15.EXE with. > Lynx and then running it; got the following: > 
 > run mpack15e1 > %DCL-W-ACTIMAGE, error activating image MPACK15i* > -CLI-E-IMGNAME, image file MPACK15.EXE;1: > -IMGACT-F-NOTNATIVE, image is not an OpenVMS Alpha image  H I'm terribly sorry. I really thought I had put up Alpha images. In fact,F these are VAX images. I'll put together an Alpha distribution over the weekend.   cu,    Martin   ------------------------------  % Date: Sat, 27 Jan 2001 06:36:10 +0100a2 From: martin@radiogaga.harz.de (Martin Vorlaender)! Subject: Re: MPACK/MUNPACK images ; Message-ID: <3a725e4a.524144494f47414741@radiogaga.harz.de>   3 Martin Vorlaender (martin@radiogaga.harz.de) wrote:eJ > I'm terribly sorry. I really thought I had put up Alpha images. In fact,H > these are VAX images. I'll put together an Alpha distribution over the
 > weekend.  B Now there are MPACK15.VAX_EXE (81 kB) and MPACK15.AXP_EXE (133 kB)8 for download at http://www.pdv-systeme.de/users/martinv/  # Again, sorry for causing confusion.k   cu,s   Martin --J One OS to rule them all       | Martin Vorlaender  |  VMS & WNT programmer7 One OS to find them           | work: mv@pdv-systeme.deaN One OS to bring them all      |       http://www.pdv-systeme.de/users/martinv/> And in the Darkness bind them.| home: martin@radiogaga.harz.de   ------------------------------  % Date: Sat, 27 Jan 2001 06:40:35 +0100n2 From: martin@radiogaga.harz.de (Martin Vorlaender)! Subject: Re: MPACK/MUNPACK imagess; Message-ID: <3a725f53.524144494f47414741@radiogaga.harz.de>o   [posted & mailed]f  3 Hoff Hoffman (hoffman@xdelta.zko.dec.nospam) wrote:nJ >   Is there have a kit (sources and pre-built images) here (or elsewhere)8 >   that can be submitted for the next OpenVMS Freeware?  G To have a clean distribution would require some work. This was a really F quick port done by Patrick Moreau of the (not quite so clean) sources.B A bit too late, I fear - IIRC, the submission deadline was Feb 5?!   cu,.   Martin --J One OS to rule them all       | Martin Vorlaender  |  VMS & WNT programmer7 One OS to find them           | work: mv@pdv-systeme.de J One OS to bring them all      |   http://www.pdv-systeme.de/users/martinv/> And in the Darkness bind them.| home: martin@radiogaga.harz.de   ------------------------------   Date: 26 Jan 2001 13:35:38 GMT1 From: bill@triangle.cs.uofs.edu (Bill Gunshannon)P* Subject: Re: New OpenVMS Times now on line, Message-ID: <94rufb$2u4s$1@info.cs.uofs.edu>  1 In article <3a6cc39e.116731991@swen.process.com>,i0  goathunter@goatley.com (Hunter Goatley) writes: |>H |> From the PDF viewer, you can select whatever printer drivers you have( |> installed: Postscript, PCL, FAX, etc. |> nG |> Yes, you can do the same thing with a .PS file and Ghostscript, but,iB |> as I noted in an earlier post, more Windows users are likely to; |> already have the Adobe Acrobat viewer installed than GS.n  B Maybe so, but remember, this whole thread started over viewing PDFB from a VAX.  I hardly think anyone here cares one way or the other' what software is available for Windows.    bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   .   ------------------------------  # Date: Fri, 26 Jan 2001 20:35:16 GMT - From: goathunter@goatley.com (Hunter Goatley)n* Subject: Re: New OpenVMS Times now on line1 Message-ID: <3a71df16.106206947@swen.process.com>i  < On 26 Jan 2001 13:35:38 GMT, bill@triangle.cs.uofs.edu (Bill Gunshannon) wrote:  2 >In article <3a6cc39e.116731991@swen.process.com>,1 > goathunter@goatley.com (Hunter Goatley) writes:? >|>lI >|> From the PDF viewer, you can select whatever printer drivers you havem) >|> installed: Postscript, PCL, FAX, etc.r >|> H >|> Yes, you can do the same thing with a .PS file and Ghostscript, but,C >|> as I noted in an earlier post, more Windows users are likely tog< >|> already have the Adobe Acrobat viewer installed than GS. >aC >Maybe so, but remember, this whole thread started over viewing PDFtC >from a VAX.  I hardly think anyone here cares one way or the otheru( >what software is available for Windows. >b@ True, but the thread had devolved into an area that did make the post relevant.   Trying to replace Carl?I   Hunter ------9 Hunter Goatley, Process Software, http://www.process.com/v9 goathunter@goatley.com     http://www.goatley.com/hunter/n   ------------------------------  % Date: Fri, 26 Jan 2001 16:20:36 -0600o* From: WILLIAM WEBB <WWEBB1@email.usps.gov>* Subject: RE: New OpenVMS Times now on line- Message-ID: <0033000014623156000002L062*@MHS>o   =0AHunter, that's bad.  
 Question1:) Is it flamage to compare someone to Carl?a  
 Question2:7 Is referencing Carl in a flame an example of recursion?u   WWWebb -----Original Message-----/ From: Info-VAX-Request@Mvb.Saic.Com at INTERNETe& Sent: Friday, January 26, 2001 4:05 PM6 To: Webb, William W; Info-VAX@Mvb.Saic.Com at INTERNET* Subject: RE: New OpenVMS Times now on line    < On 26 Jan 2001 13:35:38 GMT, bill@triangle.cs.uofs.edu (Bill Gunshannon) wrote:  2 >In article <3a6cc39e.116731991@swen.process.com>,1 > goathunter@goatley.com (Hunter Goatley) writes:a >|> H >|> From the PDF viewer, you can select whatever printer drivers you ha= ve) >|> installed: Postscript, PCL, FAX, etc.e >|> H >|> Yes, you can do the same thing with a .PS file and Ghostscript, but= ,sC >|> as I noted in an earlier post, more Windows users are likely toS< >|> already have the Adobe Acrobat viewer installed than GS. >rC >Maybe so, but remember, this whole thread started over viewing PDF C >from a VAX.  I hardly think anyone here cares one way or the other ( >what software is available for Windows. >y@ True, but the thread had devolved into an area that did make the post relevant.   Trying to replace Carl?    Hunter ------9 Hunter Goatley, Process Software, http://www.process.com/u: goathunter@goatley.com     http://www.goatley.com/hunter/=   ------------------------------  # Date: Fri, 26 Jan 2001 22:47:06 GMTs- From: goathunter@goatley.com (Hunter Goatley)a* Subject: Re: New OpenVMS Times now on line1 Message-ID: <3a71fd7d.113989698@swen.process.com>r  0 On Fri, 26 Jan 2001 16:20:36 -0600, WILLIAM WEBB <WWEBB1@email.usps.gov> wrote:   >Hunter, that's bad. >> >Question1:h* >Is it flamage to compare someone to Carl? >c >Question2:c8 >Is referencing Carl in a flame an example of recursion? >rC I wasn't flaming, just wondering why Bill had to criticize my post,cB taking it somewhat out of context.  It was relevant to the thread.@ If you have problem with the thread, criticize the thread, don't& make me out to be someone like Andrew.   Hunter ------9 Hunter Goatley, Process Software, http://www.process.com/ 9 goathunter@goatley.com     http://www.goatley.com/hunter/b   ------------------------------  # Date: Fri, 26 Jan 2001 09:18:08 GMT  From: Didier.Morandi@gmx.frb+ Subject: Re: Printing problems in TCPIP (2)e) Message-ID: <94rfcc$9a0$1@nnrp1.deja.com>b  3 In article <002101c08720$04fbfbe0$d2010001@desenv>,n4   "Valdemir J. Santos" <valdemir-@uol.com.br> wrote:E >  I'd like to get more information about using TCPIP in VMS to print * >  jobs. What Compaq manual should I get ?  < DIGITAL TCP/IP Services for OpenVMS Management, chap 18 3.2H www.openvms.compaq.com:8000/72final/6526/6526profile_028.html#lprsetup_s ec  G >  Is there any site in the Internet where I can get more knowledgementl< >  to use and configurate server queues in TCPIP under VMS ?   NO!    :-)i   D.     Sent via Deja.com0 http://www.deja.com/   ------------------------------  % Date: Fri, 26 Jan 2001 18:04:01 +0100S, From: "Bart Zorn" <B.Zorn@TrueBit.nospam.nl>% Subject: Re: Problem with decnet copy.* Message-ID: <94samb$kp6$1@buty.wanadoo.nl>  D You said "vax4 and vax6 are clustered via ethernet". Do you mean VMS: clustered? In that case, why would you want to use DECnet?  	 Bart Zornt  G <manser@decus.de> wrote in message news:009F6B43.C4F8E527.1@decus.de...a > Hi VMS Folks,t >  > ? > i am running on a vax 4705A (VAX6) openVMS 6.2 decnet osi 6.3o@ >                   MicroVAX 3100 M98 openVms 6.2 decnet osi 6.3 >r* > vax4 and vax6 are clustered via ethernet >aG > I have the following problem when copying files between vax6 and vax4i >r >l4 > VAX6> copy keypad.com vax4::$3$dka200:<manser.tmp>L > COPY DSA2:[MANSER]KEYPAD.COM;1 to VAX4::$3$DKA200:[MANSER.TMP]KEYPAD.COM;1 ? [N]: yL > %COPY-E-OPENOUT, error opening VAX4::$3$DKA200:[MANSER.TMP]KEYPAD.COM;1 as output$ > -RMS-E-CRE, ACP file create failed> > -SYSTEM-F-INVLOGIN, login information invalid at remote node9 > %COPY-W-NOTCOPIED, DSA2:[MANSER]KEYPAD.COM;1 not copiedP >s >  > VAX4> : > %%%%%%%%%%%  OPCOM  26-JAN-2001 16:47:22.33  %%%%%%%%%%%( > Message from user AUDIT$SERVER on VAX4I > Security alarm (SECURITY) and security audit (SECURITY) on VAX4, system  id: 10281 > Auditable event:          Network login failurea3 > Event time:               26-JAN-2001 16:47:22.32o$ > PID:                      20400214# > Process name:             NET$ACP9( > Username:                 DNA$SessCtrl0 > Remote node id:           00000000000000000000' > Remote node fullname:     LOCAL:.VAX64" > Remote username:          MANSERI > Status:                   %LOGIN-F-NOTVALID, user authorization failure2 >u  > 1) DECNET Proxies are enabled: > * > VAX4> mc authorize show /proxy *::manser >a' >  Default proxies are flagged with (D)c >s > LOCAL:.VAX6::MANSERo >     MANSER >a > LOCAL:.VAX7::MANSERk >     MANSER >C > LOCAL:.VAX5::MANSER- >     MANSER > * > VAX6> mc authorize show /proxy *::manser >t' >  Default proxies are flagged with (D)  >G > LOCAL:.VAX6::MANSERo >     MANSER (D) >h > LOCAL:.VAX4::MANSER  >     MANSER (D) >c > LOCAL:.VAX5::MANSERs >     MANSER (D) >o2 > VAX6> mc ncl show session control incoming proxy >m > Node 0 Session Control& > at 2001-01-26-17:02:30.810+01:00Iinf >A > Characteristicss > . >     Incoming Proxy                    = True >e2 > VAX6> mc ncl show session control outgoing proxy >e > Node 0 Session Control& > at 2001-01-26-17:02:42.570+01:00Iinf >. > Characteristics  >e. >     Outgoing Proxy                    = True > 3 > VAX4>  mc ncl show session control incoming proxys >o > Node 0 Session Control& > at 2001-01-26-17:03:25.150+01:00Iinf >a > Characteristicsu >p. >     Incoming Proxy                    = True > 3 > VAX4>  mc ncl show session control outgoing proxya >) > Node 0 Session Control& > at 2001-01-26-17:03:38.620+01:00Iinf >d > Characteristicsc >s. >     Outgoing Proxy                    = True >e' > VAX4> mc authorize show [376,*] /brietF >        Owner         Username           UIC       Account  Privs Pri	 Directoryg >.F > VPM$SERVER Default   VPM$SERVER      [376,370]    DECNETV  Normal  4 SYS$SPECIFIC:[VPM$SERVER]~F > NML$SERVER DEFAULT   NML$SERVER      [376,371]    DECNET   Normal  4 SYS$SPECIFIC:[NML$SERVER](F > FAL$SERVER DEFAULT   FAL$SERVER      [376,373]    DECNET   Normal  4 SYS$SPECIFIC:[FAL$SERVER]rF > DECNET DEFAULT       DECNET          [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]s >o' > VAX6> mc authorize show [376,*] /brieOF >        Owner         Username           UIC       Account  Privs Pri	 DirectoryS >mF > DTSS SERVER          DTSS$SERVER     [376,300]    DECNET   Normal  4 SYS$SPECIFIC:[DTSS$SERVER]F > CML$SERVER Default   CML$SERVER      [376,366]    DECNETV  Normal  4 SYS$SPECIFIC:[CML$SERVER] F > MIRRO$SERVER Default MIRRO$SERVER    [376,367]    DECNETV  Normal  4 SYS$SPECIFIC:[MIRRO$SERVER] F > VPM$SERVER Default   VPM$SERVER      [376,370]    DECNETV  Normal  4 SYS$SPECIFIC:[VPM$SERVER]eF > NML$SERVER DEFAULT   NML$SERVER      [376,371]    DECNET   Normal  4 SYS$SPECIFIC:[NML$SERVER] F > PHONE$SERVER Default PHONE$SERVER    [376,372]    DECNETV  Normal  4 SYS$SPECIFIC:[PHONE$SERVER] F > FAL$SERVER DEFAULT   FAL$SERVER      [376,373]    DECNET   Normal  4 SYS$SPECIFIC:[FAL$SERVER]cF > MAIL$SERVER Default  MAIL$SERVER     [376,374]    DECNETV  Normal  4 SYS$SPECIFIC:[MAIL$SERVER]F > DECNET DEFAULT       DECNET          [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]:F > DECNET DEFAULT       PSIMAIL         [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]hF >                      RDB$REMOTE      [376,376]             Normal  4 SYS$COMMON:[RDB$REMOTE]tF >                      SQLSRV$SRV      [376,377]             All     4 SYS$SYSDEVICE:[SQLSRV$SERVER]e >  >s >o& > could anyone tell me what's going on > thanks in advanced >b > Nazim Manser >g > email: nmanser@progis.de; >        Nazim.Manser@socgen.com work (tuesdays to fridays)'   ------------------------------  % Date: Fri, 26 Jan 2001 14:08:50 -0500s& From: "Syltrem" <syltrem@videotron.ca>% Subject: Re: Problem with decnet copy,8 Message-ID: <LTjc6.10053$Ee5.63117@wagner.videotron.net>  K If you call a menu (or something else that only works interactively) in therI login.com of MANSER, then you must make you don't call it when in network- mode. Place:% $ IF F$MODE().EQS."NETWORK" THEN EXIT0 before calling the menu.  I Other than that, check for pwd expiration, intrusion detection. Since you6G don't supply user/pwd, you should check that a proxy is enabled for thec  acount running the copy command.   Syltremh  / <manser@decus.de> a crit dans le message news:0 009F6B43.C4F8E527.1@decus.de...s > Hi VMS Folks,2 >@ >s? > i am running on a vax 4705A (VAX6) openVMS 6.2 decnet osi 6.3t@ >                   MicroVAX 3100 M98 openVms 6.2 decnet osi 6.3 > * > vax4 and vax6 are clustered via ethernet >IG > I have the following problem when copying files between vax6 and vax4t >o >44 > VAX6> copy keypad.com vax4::$3$dka200:<manser.tmp>L > COPY DSA2:[MANSER]KEYPAD.COM;1 to VAX4::$3$DKA200:[MANSER.TMP]KEYPAD.COM;1 ? [N]: yL > %COPY-E-OPENOUT, error opening VAX4::$3$DKA200:[MANSER.TMP]KEYPAD.COM;1 as output$ > -RMS-E-CRE, ACP file create failed> > -SYSTEM-F-INVLOGIN, login information invalid at remote node9 > %COPY-W-NOTCOPIED, DSA2:[MANSER]KEYPAD.COM;1 not copiedg >D >3 > VAX4>0: > %%%%%%%%%%%  OPCOM  26-JAN-2001 16:47:22.33  %%%%%%%%%%%( > Message from user AUDIT$SERVER on VAX4I > Security alarm (SECURITY) and security audit (SECURITY) on VAX4, system  id: 10281 > Auditable event:          Network login failure 3 > Event time:               26-JAN-2001 16:47:22.32 $ > PID:                      20400214# > Process name:             NET$ACPa( > Username:                 DNA$SessCtrl0 > Remote node id:           00000000000000000000' > Remote node fullname:     LOCAL:.VAX6?" > Remote username:          MANSERI > Status:                   %LOGIN-F-NOTVALID, user authorization failure. >   > 1) DECNET Proxies are enabled: >s* > VAX4> mc authorize show /proxy *::manser >h' >  Default proxies are flagged with (D)d >D > LOCAL:.VAX6::MANSERe >     MANSER >  > LOCAL:.VAX7::MANSERD >     MANSER >o > LOCAL:.VAX5::MANSERt >     MANSER >n* > VAX6> mc authorize show /proxy *::manser >s' >  Default proxies are flagged with (D)  >  > LOCAL:.VAX6::MANSERc >     MANSER (D) >  > LOCAL:.VAX4::MANSERd >     MANSER (D) >  > LOCAL:.VAX5::MANSERa >     MANSER (D) >a2 > VAX6> mc ncl show session control incoming proxy >o > Node 0 Session Control& > at 2001-01-26-17:02:30.810+01:00Iinf >i > Characteristics, >d. >     Incoming Proxy                    = True >r2 > VAX6> mc ncl show session control outgoing proxy >i > Node 0 Session Control& > at 2001-01-26-17:02:42.570+01:00Iinf >_ > Characteristicsn >_. >     Outgoing Proxy                    = True >-3 > VAX4>  mc ncl show session control incoming proxy  >  > Node 0 Session Control& > at 2001-01-26-17:03:25.150+01:00Iinf >\ > Characteristicsa >x. >     Incoming Proxy                    = True >r3 > VAX4>  mc ncl show session control outgoing proxy5 >  > Node 0 Session Control& > at 2001-01-26-17:03:38.620+01:00Iinf >0 > Characteristicsr >o. >     Outgoing Proxy                    = True >K' > VAX4> mc authorize show [376,*] /brie2F >        Owner         Username           UIC       Account  Privs Pri	 Directoryc >kF > VPM$SERVER Default   VPM$SERVER      [376,370]    DECNETV  Normal  4 SYS$SPECIFIC:[VPM$SERVER]tF > NML$SERVER DEFAULT   NML$SERVER      [376,371]    DECNET   Normal  4 SYS$SPECIFIC:[NML$SERVER]PF > FAL$SERVER DEFAULT   FAL$SERVER      [376,373]    DECNET   Normal  4 SYS$SPECIFIC:[FAL$SERVER]AF > DECNET DEFAULT       DECNET          [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]r >A' > VAX6> mc authorize show [376,*] /brierF >        Owner         Username           UIC       Account  Privs Pri	 Directorya > F > DTSS SERVER          DTSS$SERVER     [376,300]    DECNET   Normal  4 SYS$SPECIFIC:[DTSS$SERVER]F > CML$SERVER Default   CML$SERVER      [376,366]    DECNETV  Normal  4 SYS$SPECIFIC:[CML$SERVER]nF > MIRRO$SERVER Default MIRRO$SERVER    [376,367]    DECNETV  Normal  4 SYS$SPECIFIC:[MIRRO$SERVER]gF > VPM$SERVER Default   VPM$SERVER      [376,370]    DECNETV  Normal  4 SYS$SPECIFIC:[VPM$SERVER] F > NML$SERVER DEFAULT   NML$SERVER      [376,371]    DECNET   Normal  4 SYS$SPECIFIC:[NML$SERVER]vF > PHONE$SERVER Default PHONE$SERVER    [376,372]    DECNETV  Normal  4 SYS$SPECIFIC:[PHONE$SERVER]uF > FAL$SERVER DEFAULT   FAL$SERVER      [376,373]    DECNET   Normal  4 SYS$SPECIFIC:[FAL$SERVER]kF > MAIL$SERVER Default  MAIL$SERVER     [376,374]    DECNETV  Normal  4 SYS$SPECIFIC:[MAIL$SERVER]F > DECNET DEFAULT       DECNET          [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]-F > DECNET DEFAULT       PSIMAIL         [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET])F >                      RDB$REMOTE      [376,376]             Normal  4 SYS$COMMON:[RDB$REMOTE]oF >                      SQLSRV$SRV      [376,377]             All     4 SYS$SYSDEVICE:[SQLSRV$SERVER]  >u >s >d& > could anyone tell me what's going on > thanks in advancet >  > Nazim Manser >S > email: nmanser@progis.de; >        Nazim.Manser@socgen.com work (tuesdays to fridays)u   ------------------------------  % Date: Fri, 26 Jan 2001 19:50:37 -0500 ) From: "C Castiglia" <ccastiglia@snet.net>d% Subject: Re: Problem with decnet copy 5 Message-ID: <zXoc6.24516$OD.9447952@typhoon.snet.net>M  E I could be wrong about this, but when I copy from AlphaServer runningdJ OpenVMS 7.1-1H2 to a Microvax running a lower version of OpenVMS (5.x?), I have to do it like this:  $    $ COPY FOO.BAR MVAX_NODE"USERNAME, PASSWORD"::DKA100:[DIRLEVL1.DIRLEVL2]FOO.BAR  K Username and password must be a valid account on the target machine.  TherejL are no spaces or other characters between the target node name and the first quote character.      G <manser@decus.de> wrote in message news:009F6B43.C4F8E527.1@decus.de...t > Hi VMS Folks,  >F > ? > i am running on a vax 4705A (VAX6) openVMS 6.2 decnet osi 6.3 @ >                   MicroVAX 3100 M98 openVms 6.2 decnet osi 6.3 >e* > vax4 and vax6 are clustered via ethernet >tG > I have the following problem when copying files between vax6 and vax4> >r >y4 > VAX6> copy keypad.com vax4::$3$dka200:<manser.tmp>L > COPY DSA2:[MANSER]KEYPAD.COM;1 to VAX4::$3$DKA200:[MANSER.TMP]KEYPAD.COM;1 ? [N]: yL > %COPY-E-OPENOUT, error opening VAX4::$3$DKA200:[MANSER.TMP]KEYPAD.COM;1 as output$ > -RMS-E-CRE, ACP file create failed> > -SYSTEM-F-INVLOGIN, login information invalid at remote node9 > %COPY-W-NOTCOPIED, DSA2:[MANSER]KEYPAD.COM;1 not copiede >  >v > VAX4> : > %%%%%%%%%%%  OPCOM  26-JAN-2001 16:47:22.33  %%%%%%%%%%%( > Message from user AUDIT$SERVER on VAX4I > Security alarm (SECURITY) and security audit (SECURITY) on VAX4, systeml id: 10281 > Auditable event:          Network login failureo3 > Event time:               26-JAN-2001 16:47:22.32c$ > PID:                      20400214# > Process name:             NET$ACPa( > Username:                 DNA$SessCtrl0 > Remote node id:           00000000000000000000' > Remote node fullname:     LOCAL:.VAX6m" > Remote username:          MANSERI > Status:                   %LOGIN-F-NOTVALID, user authorization failure  >   > 1) DECNET Proxies are enabled: >w* > VAX4> mc authorize show /proxy *::manser >n' >  Default proxies are flagged with (D)r >r > LOCAL:.VAX6::MANSERd >     MANSER >A > LOCAL:.VAX7::MANSER  >     MANSER >y > LOCAL:.VAX5::MANSERi >     MANSER >a* > VAX6> mc authorize show /proxy *::manser >t' >  Default proxies are flagged with (D)e >  > LOCAL:.VAX6::MANSERa >     MANSER (D) >  > LOCAL:.VAX4::MANSERd >     MANSER (D) >a > LOCAL:.VAX5::MANSERt >     MANSER (D) >t2 > VAX6> mc ncl show session control incoming proxy >c > Node 0 Session Control& > at 2001-01-26-17:02:30.810+01:00Iinf >t > Characteristics/ >t. >     Incoming Proxy                    = True >,2 > VAX6> mc ncl show session control outgoing proxy >1 > Node 0 Session Control& > at 2001-01-26-17:02:42.570+01:00Iinf >- > Characteristics5 >0. >     Outgoing Proxy                    = True >n3 > VAX4>  mc ncl show session control incoming proxys >n > Node 0 Session Control& > at 2001-01-26-17:03:25.150+01:00Iinf >b > Characteristicss >-. >     Incoming Proxy                    = True >E3 > VAX4>  mc ncl show session control outgoing proxy  >l > Node 0 Session Control& > at 2001-01-26-17:03:38.620+01:00Iinf >m > Characteristics  >2. >     Outgoing Proxy                    = True >(' > VAX4> mc authorize show [376,*] /brie3F >        Owner         Username           UIC       Account  Privs Pri	 Directoryt >aF > VPM$SERVER Default   VPM$SERVER      [376,370]    DECNETV  Normal  4 SYS$SPECIFIC:[VPM$SERVER]:F > NML$SERVER DEFAULT   NML$SERVER      [376,371]    DECNET   Normal  4 SYS$SPECIFIC:[NML$SERVER]iF > FAL$SERVER DEFAULT   FAL$SERVER      [376,373]    DECNET   Normal  4 SYS$SPECIFIC:[FAL$SERVER]eF > DECNET DEFAULT       DECNET          [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]a >d' > VAX6> mc authorize show [376,*] /brieyF >        Owner         Username           UIC       Account  Privs Pri	 Directoryd >.F > DTSS SERVER          DTSS$SERVER     [376,300]    DECNET   Normal  4 SYS$SPECIFIC:[DTSS$SERVER]F > CML$SERVER Default   CML$SERVER      [376,366]    DECNETV  Normal  4 SYS$SPECIFIC:[CML$SERVER]uF > MIRRO$SERVER Default MIRRO$SERVER    [376,367]    DECNETV  Normal  4 SYS$SPECIFIC:[MIRRO$SERVER]2F > VPM$SERVER Default   VPM$SERVER      [376,370]    DECNETV  Normal  4 SYS$SPECIFIC:[VPM$SERVER]TF > NML$SERVER DEFAULT   NML$SERVER      [376,371]    DECNET   Normal  4 SYS$SPECIFIC:[NML$SERVER]6F > PHONE$SERVER Default PHONE$SERVER    [376,372]    DECNETV  Normal  4 SYS$SPECIFIC:[PHONE$SERVER]tF > FAL$SERVER DEFAULT   FAL$SERVER      [376,373]    DECNET   Normal  4 SYS$SPECIFIC:[FAL$SERVER]oF > MAIL$SERVER Default  MAIL$SERVER     [376,374]    DECNETV  Normal  4 SYS$SPECIFIC:[MAIL$SERVER]F > DECNET DEFAULT       DECNET          [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]iF > DECNET DEFAULT       PSIMAIL         [376,376]    DECNET   Normal  4 SYS$SPECIFIC:[DECNET]lF >                      RDB$REMOTE      [376,376]             Normal  4 SYS$COMMON:[RDB$REMOTE]/F >                      SQLSRV$SRV      [376,377]             All     4 SYS$SYSDEVICE:[SQLSRV$SERVER]i >x >b >u& > could anyone tell me what's going on > thanks in advancec >a > Nazim Manser >b > email: nmanser@progis.de; >        Nazim.Manser@socgen.com work (tuesdays to fridays)r   ------------------------------  % Date: Fri, 26 Jan 2001 00:20:26 -0800 + From: "Wayne Holland" <wholland@tscnet.com> : Subject: Re: Question: Info. Needed On SHD51-ZZ Drive Box.O Message-ID: <327F1A4F997C0E09.F8CCE9F60146E3A0.2C9B6473E9A2F1DA@lp.airnews.net>#  4 "Ben Myers" <benmyers@ultranet.com> wrote in message. news:3a7102e4.59057007@news.ma.ultranet.com...H > Nice gift!  Why not open up one of the canisters to see what's inside.F > At least you'll know the capacity of the drives you're dealing with. >- > ... Ben Myers  >    OH! Come on! At least help him out!  5 > anon314@mail.ourservers.net (Anonymous #314) wrote:n >c > > I > >I have a drive chassis box that I've been given and I'm trying to findl > >information on it.D > > I > >It says it's SHD51-ZZ and it's full of drive canisters (with drives int them). > >nJ > >I would appreciate any information or links to information on this box. > >lA > >I've tried search the Compaq/DEC site and true to form all thee information 7 > >that I used to rely on is gone, true to Compaq form.i > >  > >Thanks in advance.  > >e >    ------------------------------   Date: 26 Jan 2001 18:53:37 GMTX From: alan@nabeth.cxo.dec.com (Alan Rollow - Dr. File System's Home for Wayward Inodes.): Subject: Re: Question: Info. Needed On SHD51-ZZ Drive Box.* Message-ID: <94sh3h$i7v@usenet.pa.dec.com>  f In article <PP7kqRrJi0cm@cartman.ourservers.net>, anon314@mail.ourservers.net (Anonymous #314) writes:  G >I have a drive chassis box that I've been given and I'm trying to findC >information on it.E >DN >It says it's SHD51-ZZ and it's full of drive canisters (with drives in them).  A I looked at Service's part number validator, and that part wasn'ty< found.  Other SHD* parts appear to be some family of storage> enclosure for 3.5" form factor disks.  This could just be some@ variety of StorageWorks enclosure, or it could be something much> older.  Some of the parts suggest this might have been part of some RAID subsystem.  E A more detailed physical description of the enclosure and part number ' of the drives (if any) would be useful.   K >I've tried search the Compaq/DEC site and true to form all the information,5 >that I used to rely on is gone, true to Compaq form.r  C This could have been an enclosure for an OEM subsystem that we solduD to somebody else.  Since it wasn't for direct sale, it wouldn't have, been in the old Systems and Options Catalog.   >  >Thanks in advance.- >  >  --& Alan Rollow				alan@nabeth.cxo.dec.com   ------------------------------  # Date: Fri, 26 Jan 2001 22:30:29 GMTX2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman): Subject: Re: Question: Info. Needed On SHD51-ZZ Drive Box.7 Message-ID: <9Umc6.143$cu.1115@gazette.loc1.tandem.com>g  } In article <327F1A4F997C0E09.F8CCE9F60146E3A0.2C9B6473E9A2F1DA@lp.airnews.net>, "Wayne Holland" <wholland@tscnet.com> writes:  : 5 :"Ben Myers" <benmyers@ultranet.com> wrote in messaget/ :news:3a7102e4.59057007@news.ma.ultranet.com... I :> Nice gift!  Why not open up one of the canisters to see what's inside.uG :> At least you'll know the capacity of the drives you're dealing with.c :r
 :OH! Come on!x :At least help him out!C    H   Um, Wayne, SHD51-ZZ is a storage shelf assembly.  Contents can vary.  @   Ben's suggestion to look at the contents is correct and valid.    N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Fri, 26 Jan 2001 23:09:16 GMT 2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)D Subject: Re: Rdb and EV6 (was: Re: difference between EV6 and EV67?)7 Message-ID: <wsnc6.146$cu.1056@gazette.loc1.tandem.com>~  F   Y'all will have to ask Oracle Rdb support for any background on the K   decision on the particular OpenVMS versions and microprocessor platforms  I   that might be available.  I'm certainly not in a position to comment...C    Y In article <3A719A5E.D6113759@bbc.co.uk>, Tim Llewellyn <tim.llewellyn@bbc.co.uk> writes:s :  :Jan Vorbrueggen wrote:O :rM :> I thought it was EV6 processors in general? The EV6 actually makes use of SE :> some of the more relaxed memory consistency features of the Alpha PF :> architecture, as I understand it, and previous compilers generated G :> broken LL/SC code that was OK on EV4 and EV5, but could break on EV6MH :> systems. DEC^H^H^HCompaq has a program to go through executables and G :> find candidates for such mistaken code sequences, but it isn't fool 4F :> proof (it can generate false positives). Apparently, Oracle wasn't C :> prepared to check and/or re-issue older versions of its two VMS V& :> database products for this feature.    I   Um, Interesting theory.  The availability of the Oracle Rdb ECOs V6.0A RJ   ECO 6 (V6.0-16), V6.1-14, and V7.0-14 (at least) would tend to indicate I   another interpretation is available, however.  These are all SRM-clean.F  J   The old GEM code generator back-end was capable of generating a sequenceI   that was prohibited by the Alpha architecture.  EV6 was the first Alpha[H   microprocessor that "cared" about this particular coding error.  OlderH   versions of Rdb included a SQL pre-processor with the erroroneous GEM.  J   I don't know the specific Rdb versions and ECOs that first included the D   new SRM-clean GEM and had any lurking non-SRM-clean code rebuilt, B   off-hand.  You'd need to check with Oracle Rdb support for this G   information.  (I know some of my Rdb-based code was not SRM-clean...)i  E   Details and tools for checking for the SRM-bogus code sequence are 5D   available, with information included in OpenVMS documentation for H   releases with EV6 support, postings here, information at the website, G   and the detection tool documentation are also available in the 21264 )9   directory of the OpenVMS Freeware V4 kit (and website).   L :So, Compaq broke VMS's backwards compatibilty, one of its' major strengths., :No wonder Oracle ran away again eventually.  8   The GEM code generator problem predates Compaq, BTW...  /   I will assume you meant upward-compatibility.i  B   The Alpha microprocessors are upward-compatible across the line.  K   You will want to discuss Oracle Rdb OpenVMS version and platform support sK   decisions with an Oracle representative.  (Again, I'm not in position...)d  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Fri, 26 Jan 2001 16:22:32 +0000.; From: Andrew Harrison SUNUK Consultancy <andrew@uk.sun.com>CY Subject: Re: Rehash of Andrew's Java on OpenVMS FUD.  (was: Re: GS160 hardware     questig) Message-ID: <3A71A448.5020306@uk.sun.com>s   Arne Vajhj wrote:   > andrew harrison wrote: > 9 >> Well its hardly bogus, its just worse than I suggested 8 >> The 1.2.X JVM is available for OpenVMS and neither is8 >> the plugin. The 1.3.X JVM isn't available for OpenVMS+ >> at all (well there is an alpha release).$ >  >   > It is indeed not good for VMS. > C > But it illustrates how well you check your facts before posting !v >   ( Incedentally you were not very quick off+ the mark to correct Cristof and Jordan when - they both claimed that having a 1.3 JVM isn'ta
 important.  4 You of course knew that this was untrue particularly/ for people who want to run Java in a browser on  OpenVMS.  ) Perhaps you should have piped up earlier.n   > Arne RegardsR Andrew Harrison    ------------------------------  % Date: Fri, 26 Jan 2001 14:50:44 -0500 ' From: "Dale A. Marcy" <dqm@y12.doe.gov>o6 Subject: Re: RMS bug, possible triggered by autoextend+ Message-ID: <3A71D514.1AFE265F@y12.doe.gov>    John,S  K      Bill's comment reminded me of a problem we saw with storage areas in a K DBMS database that were shadowed between two buildings.  When verifying thesM database, sometimes we would get an error on a database area and sometimes weAI would not.  Through testing, I discovered that if the verify was run on aRJ machine in building A, it would succeed, but if it was run on a machine inH building B it would fail.  We determined that it was a shadowing problemM through conversation with support and applied a shadowing ECO and the problemoJ never resurfaced (it was happening every 2 or 3 months before the patch). N This was a few years ago, so it was a different version of VMS, I think it wasG V6.2 and we had a 4 node VAX cluster at that time, with 2 VAXes in each 	 building.   
 Dale A. MarcyC. Science Applications International Corporation   Bill Todd wrote: > / > John Nebel <nebel@csdco.com> wrote in messageeC > news:Pine.OSF.4.21.0101260952540.17891-100000@athena.csdco.com...e > >s# > > Wonder if anyone has seen this:8 > >0K > > a relative file becomes corrupted - this happened in several cases, one N > > was definitely after an autoextend - so that ana/rms reports the following > > 0 > > ***  VBN 69602:  Reserved flag bit 0 is set.0 > > ***  VBN 69602:  Reserved flag bit 4 is set.0 > > ***  VBN 69602:  Reserved flag bit 5 is set. > > J > > Successive executions of ana/rms sometimes report the error, sometimes4 > > don't even though the file hasn't been modified. > L > Not that I have a clue what I'm talking about, but doesn't that sound moreK > like a shadowing bug, where sometimes you read the bad copy and sometimesN > you don't? >  > - bill >  > >3L > > Mixed version cluster, VAX/VMS 7.2, AXP/VMS 7.2-1.  The disk is a 3-copy1 > > shadow set with one remote (over a DS3) copy.$ > >R, > > The autoextend was performed by VAX/VMS. > >[I > > Haven't heard back from support and thought I'd ask here just in caseN > > anyone has seen it.M > >V > > Thanks.$ > >R > > John Nebel > >    ------------------------------  % Date: Fri, 26 Jan 2001 14:14:41 -0500V' From: "Bill Todd" <billtodd@foo.mv.com> 6 Subject: Re: RMS bug, possible triggered by autoextend( Message-ID: <94si2t$kj0$1@pyrite.mv.net>  - John Nebel <nebel@csdco.com> wrote in message A news:Pine.OSF.4.21.0101260952540.17891-100000@athena.csdco.com...  > ! > Wonder if anyone has seen this:$ >VI > a relative file becomes corrupted - this happened in several cases, one[L > was definitely after an autoextend - so that ana/rms reports the following > . > ***  VBN 69602:  Reserved flag bit 0 is set.. > ***  VBN 69602:  Reserved flag bit 4 is set.. > ***  VBN 69602:  Reserved flag bit 5 is set. > H > Successive executions of ana/rms sometimes report the error, sometimes2 > don't even though the file hasn't been modified.  J Not that I have a clue what I'm talking about, but doesn't that sound moreI like a shadowing bug, where sometimes you read the bad copy and sometimesu
 you don't?   - bill   >lJ > Mixed version cluster, VAX/VMS 7.2, AXP/VMS 7.2-1.  The disk is a 3-copy/ > shadow set with one remote (over a DS3) copy.c >w* > The autoextend was performed by VAX/VMS. >-G > Haven't heard back from support and thought I'd ask here just in casea > anyone has seen it.d >u	 > Thanks.P >l > John Nebel >p   ------------------------------   Date: 26 Jan 2001 14:38:24 GMT7 From: Thomas.Hahnemann@nospam_s-t.de (Thomas Hahnemann)u Subject: Re: samba 2.0.60 Message-ID: <Oozvf8elmJpy-pn2-SgUMd7m4K7Qu@Tom2>   John,a  A thanks for your help. Now I've got MMK and can compile FRONTPORT.O   dka300:[frontport] mmk /DESCRIP=FRONPORTr  E Fist break is on the source file fport__getset_samba_mode.c line 225. ? there is a fab files l_xab which seems to be dclared as char *.d  = There are a few informational messages about unused includes.s  B After all is compiled, link warns there are unresolved externals : 	DLCLOSE 	DLERROR 	DLOPEN  	DLSYM  4 The object fport__dlopen.obj_axp was properly build.   Did I something wrong ?t   Regardse     Thomas Hahnemann   ------------------------------  % Date: Fri, 26 Jan 2001 12:12:14 -0500e% From: "Brian Tillman" <tillman_brian>6 Subject: Re: samba 2.0.6$ Message-ID: <3a71af08$1@news.si.com>  9 >$defcon := define/system/executive/translation=concealed   > And, of course, the /EXECUTIVE_MODE qualifier is not required.L /EXECUTIVE_MODE's purpose is for apps that need to trust that a logical nameE was defined by a systems administrator, like device logical names andE SYSUAF.M --A Brian Tillman                   Internet: tillman_brian at si.com A Smiths Aerospace                          tillman at swdev.si.comV= 3290 Patterson Ave. SE, MS      Addresses modified to preventU< Grand Rapids, MI 49512-1991     SPAM.  Replace "at" with "@"8        This opinion doesn't represent that of my company   ------------------------------    Date: 26 Jan 2001 14:06:41 -05002 From: malmberg@eisner.decus.org (John E. Malmberg) Subject: Re: samba 2.0.6+ Message-ID: <I6UiI9oI4hLW@eisner.decus.org>o  0 In article <Oozvf8elmJpy-pn2-SgUMd7m4K7Qu@Tom2>,9 Thomas.Hahnemann@nospam_s-t.de (Thomas Hahnemann) writes:T > John,e >uC > thanks for your help. Now I've got MMK and can compile FRONTPORT.4 >c > dka300:[frontport] > mmk /DESCRIP=FRONPORTt >oG > Fist break is on the source file fport__getset_samba_mode.c line 225. A > there is a fab files l_xab which seems to be dclared as char *.   G I wil have to look into that one.  It compiles cleanly on OpenVMS ALPHAp@ 7.2 and OpenVMS VAX 7.1 with the Compaq C 6.2 compiler.  I think8 I have it set to /WARN=(ENABLE=4,QUESTCODE) to be picky.  7 Can you post the exact diagnostic that you are getting?n  ? > There are a few informational messages about unused includes.l > D > After all is compiled, link warns there are unresolved externals :
 > 	DLCLOSE
 > 	DLERROR	 > 	DLOPEN  > 	DLSYM > 6 > The object fport__dlopen.obj_axp was properly build.   > Did I something wrong ?2  E No, OpenVMS 7.1 does not have those functions.  SAMBA 2.0.6 also doesi not use those functions.  = I have them conditionally compiled out on the VAX plaform ands9 should have made it determined by the version of OpenVMS.0  > If you examine the source fport__dlopen.c you well see an easy; edit to eliminate the link warnings.  It of course will notl make them work.   C I intended to rewrite them using LIB$FIND_IMAGE_SYMBOL so that theym were not version dependent.    To roll your own:        dlopen():         malloc a structure and stick the filename into it.  
     dlerror()e=         returns a value from an error variable shared between          the routines.M  >     dlsym() uses the structure that was malloc() by the dlopen/ 		replacement, and calls LIB$FIND_IMAGE_SYMBOL.7  
     dlclose() ) 	frees the structure malloc() by dlopen()A   -JohnD wb8tyw@qsl.network   ------------------------------   Date: 26 Jan 2001 21:48:20 GMT7 From: Thomas.Hahnemann@nospam_s-t.de (Thomas Hahnemann)  Subject: Re: samba 2.0.60 Message-ID: <Oozvf8elmJpy-pn2-YTIHEoGMLBCG@Tom2>  # Now it compiles and link frontport.   3 I've unpacked SAMBA_VMS_SOURCE_2_0_6_JEM_1.ZIP over 2 SAMBA_SOURCE_2_0_6.ZIP. It compiles. But it breaks during linking :   Linking LCL_BIN:UBIQX.EXE_AXP$@ LINK /share=LCL_BIN:UBIQX.EXE_AXP  smb_ubiqx:ubiqx_AXP.opt/opt,  lcl_bin:ubiqx.OLB_AXP/lib H %LINK-F-OPENIN, error opening DATEN2:[SAMBA.SOURCE.BIN]UBIQX.OLB_AXP; as inputF -RMS-E-FNF, file not foundE %MMK-F-ERRUPD, error status %X1064109C occurred when updating target C LCL_BIN:UBIQX.EXE_AXPM  8 I've looked into the samba.mms but I could not find any # command for building UBIQX.OLB_AXP    ; In samba.mms I've added /WARN=DISABLE(UNUSEDTOP) to disablea- the often appearing "unused include" warning.R   RegardsF     Thomas Hahnemann   ------------------------------    Date: 26 Jan 2001 19:22:04 -05002 From: malmberg@eisner.decus.org (John E. Malmberg) Subject: Re: samba 2.0.6+ Message-ID: <rBjxLYtza+HT@eisner.decus.org>[  0 In article <Oozvf8elmJpy-pn2-YTIHEoGMLBCG@Tom2>,9 Thomas.Hahnemann@nospam_s-t.de (Thomas Hahnemann) writes:N% > Now it compiles and link frontport.  > 5 > I've unpacked SAMBA_VMS_SOURCE_2_0_6_JEM_1.ZIP over 4 > SAMBA_SOURCE_2_0_6.ZIP. It compiles. But it breaks > during linking : >[ > Linking LCL_BIN:UBIQX.EXE_AXPYA > LINK /share=LCL_BIN:UBIQX.EXE_AXP  smb_ubiqx:ubiqx_AXP.opt/opt,a > lcl_bin:ubiqx.OLB_AXP/libaJ > %LINK-F-OPENIN, error opening DATEN2:[SAMBA.SOURCE.BIN]UBIQX.OLB_AXP; as > inputn > -RMS-E-FNF, file not foundF > %MMK-F-ERRUPD, error status %X1064109C occurred when updating target > LCL_BIN:UBIQX.EXE_AXPa > 9 > I've looked into the samba.mms but I could not find any $ > command for building UBIQX.OLB_AXP  A Because SAMBA has so many files in each directory, I replaced the ? "default" rule for compiling C programs with one that will alsoe? put them in an object library that is named after the directory @ the source module was found in.  The object library is stored in lcl_bin:  E If I had not done that, the SAMBA.MMS rules would have been extremelyoB large.  If you think they are bad now, just think of what it wouldC have been if I had had to specify a compiler action and a librarianv* action individually for each of the files.  D I left the other UN*X stuff in the MMS file because it does not hurt@ and I thought it might make it easier to map to the order things: are in are in the UN*X makefile source for later versions.    A To make this a rule based on the file name, the command proceduremE SMB_SRC:add_to_library is run after the compile to parse out the path " that the source module came in on.    A Because it is a multiple action rule, be aware that if there is a B compiler warning and MMS or MMK aborts, the object module will notC get put in the library.  And if the library does not exist, it will  not get recreated.    C Information messages from the Compiler will not stop the build, norr6 will they prevent the object library from being built.  A The rule can be found by searching for ".c.$(O)" in the SAMBA.MMSn file.n    @ For the few C modules that were not to be in object libraries, I@ specified the action instead of letting it default to adding the object module to a library.     = So the dependencies for lcl_bin:ubiqx.$(EXE) are dependent ons? $(UBIQX_OBJ) and $(UBIQX_OBJ) expands out to be the list of all  of the source modules.  ? I did not use a separate rule for inserting an object module in ? a library if it was newer than the one that was in the library. & That has never worked reliably for me.    > I would recommend deleting all lcl_ubiqx:*.obj_axp;* files and0 running the MMK/DESCRIP=SMB_SRC:SAMBA.MMS again.  ; Something appears to have prevented all actions being takens when the compile is done.r  = > In samba.mms I've added /WARN=DISABLE(UNUSEDTOP) to disablee/ > the often appearing "unused include" warning.r  ; If that is a WARNING and not an Informational message, that ; difference in the C compiler (I did not apply any ECOs) you	B are using and the one I used could account for your build problem.  < Any module that compiled with a warning and thus stopped the: SAMBA.MMS procedure from continuing will usually need it's> corresponding object file deleted.  You will find all the ones5 that this happened to when you get to the link phase.0  < The longer term solution would probably to add the following line to the config.h file.  ! #pragma message disable unusedtop>   But your solution is good too.  B Since all of SAMBA includes a file includes.h that includes almost> all the possible header files, it is verly likely that some of7 the include files are not needed for a specific module.O   -Johnn wb8tyw@qsl.network   ------------------------------  # Date: Fri, 26 Jan 2001 14:47:24 GMTs# From: marinto <marinto@my-deja.com>o% Subject: select() and getservbyname().) Message-ID: <94s2lp$ofi$1@nnrp1.deja.com>s  F I've written some tcpip routines for unix and am about to port them toC vms using system services. After a quick look in the table "OpenVMS-; System Service and Equivalent C Socket Call" in the OpenVMSaG documentation I missed two functions that I need. The two functions are: select() and getservbyname().7   Thanks in advance.   /Marinto     Sent via Deja.comm http://www.deja.com/   ------------------------------   Date: 26 Jan 2001 09:49 CSTn' From: carl@gerg.tamu.edu (Carl Perkins)i) Subject: Re: select() and getservbyname()v- Message-ID: <26JAN200109495932@gerg.tamu.edu>m  ' marinto <marinto@my-deja.com> writes...@G }I've written some tcpip routines for unix and am about to port them torD }vms using system services. After a quick look in the table "OpenVMS< }System Service and Equivalent C Socket Call" in the OpenVMSH }documentation I missed two functions that I need. The two functions are }select() and getservbyname(). }  }Thanks in advance.o } 	 }/Marintoo  H The getservbyname() function is present in the C libraries, so you don't= need to replace it with anything else. From the on-line help:c   CC     Socket_Routinesb       getservbyname>    A            Gets information on the named service from the network>E            services database. It searches the services database untildK            a matching service name is found or until an EOF is encountered.b  F            If a non-null protocol name is also supplied, searches must#            also match the protocol.               Syntax:  "                 #include <netdb.h>  G                 struct servent *getservbyname(char *name, char *proto);n  J On the other hand, select() isn't (at least not in my somewhat old versionJ of C). It isn't the "VMS way" of doing things. Others may have suggestionsJ as to how to work around this, but it probably depends on exactly what you are trying to do.    --- Carl   ------------------------------   Date: 26 Jan 2001 17:32:26 GMT2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)) Subject: Re: select() and getservbyname()w, Message-ID: <94scba$b7o@gap.cco.caltech.edu>  O In article <94s2lp$ofi$1@nnrp1.deja.com>, marinto <marinto@my-deja.com> writes: G >I've written some tcpip routines for unix and am about to port them tooD >vms using system services. After a quick look in the table "OpenVMS< >System Service and Equivalent C Socket Call" in the OpenVMSH >documentation I missed two functions that I need. The two functions are >select() and getservbyname(). >n >Thanks in advance.C >a   Have a look at:a  9   http://seqaxp.bio.caltech.edu/www/UNIX_TO_VMS_NOTES.TXTa5   http://seqaxp.bio.caltech.edu/www/X11_VMS_NOTES.TXTt   David Mathog mathog@seqaxp.bio.caltech.edua? Manager, sequence analysis facility, biology division, Caltech s   ------------------------------  % Date: Fri, 26 Jan 2001 11:12:51 -0800a! From: Shane.F.Smith@Healthnet.comm$ Subject: Re: SGI x Itanium x ClusterD Message-ID: <OF027D61C4.AD81E95A-ON882569E0.006983F6@foundation.com>  5 Hmmm, no benchmark figures or meaningful comparisons.-   ShaneH          = fabio_compaq@ep-bc.petrobras.com.br on 01/26/2001 09:16:25 AM    To:   Info-VAX@Mvb.Saic.Com- cc:a  ! Subject:  SGI x Itanium x Cluster      Click at    A http://www.sgi.com/newsroom/press_releases/2001/january/ohio.htmlA     Regardso   FC   ------------------------------    Date: 26 Jan 2001 12:17:22 -0500* From: young_r@eisner.decus.org (Rob Young)$ Subject: Re: SGI x Itanium x Cluster+ Message-ID: <FdGxHketcMMm@eisner.decus.org>g  x In article <OF5C6E3780.3F2EE005-ON032569E0.005C123D@ep-bc.petrobras.com.br>, fabio_compaq@ep-bc.petrobras.com.br writes:
 > Click at >  > C > http://www.sgi.com/newsroom/press_releases/2001/january/ohio.html  >  >    	Fabio,t  / 	Do you have hair like Fabio?  Couldn't resist. 
 	Seriously...l  ; 	I haven't got time to prove this (but will basically state < 	it is provable... my memory is flakey at times as I am over< 	40 but like anyone if you REALLY wish to memorize something> 	or recall it, you can generally get the job done).  With that= 	as a tangential introduction, you will always see a positivedE 	Itanium story near or preceding a BAD Itanium story.  As an example, : 	about a year or so ago (prior to the first revelation of @ 	an Itanium delay) SAS announced something similar to the above.> 	The Intel trade-write lapdogs did what a lapdog does and they 	lapped it up (pun intended) !!l  ? 	This go round?  Guess what?  We now hear that it will probablyd: 	take two more steppings (at least one) to have Itanium go< 	production.  B3 was *supposed* to be a production part and > 	March was the "go live" so they could deliver on the new/mostA 	recent *projected* hardware deliverable date of first-half 2001.w  ? 	Sadly (or gleefully) first-half 2001 is most likely a far gonea< 	deal.  It ain't gonna happen.  But never mind Itanium, just) 	wait until you get a load of McKinley!!!e   	Oh... you can use:t   http://www.aceshardware.com/  ? 	as a starting place to hear early rumblings of Itanium delays.r  ; 	Early is the keyword!  But you can almost guess that IntelX< 	will let something out on a late Friday basically admitting	 	as much.C   				Robp   ------------------------------  % Date: Fri, 26 Jan 2001 20:01:07 +0000<% From: Alan Greig <a.greig@virgin.net>e$ Subject: Re: SGI x Itanium x Cluster* Message-ID: <3A71D783.8CE9DDE6@virgin.net>  " Shane.F.Smith@Healthnet.com wrote:  7 > Hmmm, no benchmark figures or meaningful comparisons.l >s > Shane  >   E Capellas mentioned that Compaq had just lost one of the supercomputergB bids after winning six in a row. Didn't say which one it was but ID guess this was it. If Compaq bid Alpha and lost that's worse than ifE they hadn't bid at all. Doesn't matter if you are best technically ifs you can't sell it. See VMS...    >'? > fabio_compaq@ep-bc.petrobras.com.br on 01/26/2001 09:16:25 AMv >  > To:   Info-VAX@Mvb.Saic.Comt > cc:t >a# > Subject:  SGI x Itanium x Cluster  >r
 > Click at > C > http://www.sgi.com/newsroom/press_releases/2001/january/ohio.html3 > 	 > Regardss >  > FC   --
 Alan Greig   ------------------------------    Date: 26 Jan 2001 15:17:43 -0500* From: young_r@eisner.decus.org (Rob Young)$ Subject: Re: SGI x Itanium x Cluster+ Message-ID: <0RxSeDVvBeKR@eisner.decus.org>   R In article <3A71D783.8CE9DDE6@virgin.net>, Alan Greig <a.greig@virgin.net> writes:    7 	The Ohio cluster isn't a supercomputer bid.  It is 142 < 	CPUs. 71 - two processor Itanium boxes in a Myrinet.  There; 	are dozens of much larger Beowulfs around the world.  What ( 	that link points to is marketing fluff.   				Robl   >  > $ > Shane.F.Smith@Healthnet.com wrote: > 8 >> Hmmm, no benchmark figures or meaningful comparisons. >> >> Shane >> > G > Capellas mentioned that Compaq had just lost one of the supercomputersD > bids after winning six in a row. Didn't say which one it was but IF > guess this was it. If Compaq bid Alpha and lost that's worse than ifG > they hadn't bid at all. Doesn't matter if you are best technically if  > you can't sell it. See VMS...  >  >>@ >> fabio_compaq@ep-bc.petrobras.com.br on 01/26/2001 09:16:25 AM >> >> To:   Info-VAX@Mvb.Saic.Com >> cc: >>$ >> Subject:  SGI x Itanium x Cluster >> >> Click att >>D >> http://www.sgi.com/newsroom/press_releases/2001/january/ohio.html >>
 >> Regards >> >> FC  >  > -- > Alan Greig >  >    ------------------------------  % Date: Fri, 26 Jan 2001 20:33:15 +0000h% From: Alan Greig <a.greig@virgin.net>o$ Subject: Re: SGI x Itanium x Cluster* Message-ID: <3A71DF0A.AC965CFC@virgin.net>   Rob Young wrote:  T > In article <3A71D783.8CE9DDE6@virgin.net>, Alan Greig <a.greig@virgin.net> writes: > @ >         The Ohio cluster isn't a supercomputer bid.  It is 142E >         CPUs. 71 - two processor Itanium boxes in a Myrinet.  There:D >         are dozens of much larger Beowulfs around the world.  What1 >         that link points to is marketing fluff.n >m >e  P Rob I don't really disagree with virtually anything you say technically. But youV either don't or won't get it: "Marketing fluff" is almost all there is. It is the mostQ important single factor. Do you really not understand this or do you think we arenC taken in? You don't have to convince us of the technical arguments.t   --
 Alan Greig   ------------------------------    Date: 26 Jan 2001 15:47:12 -0500* From: young_r@eisner.decus.org (Rob Young)$ Subject: Re: SGI x Itanium x Cluster+ Message-ID: <V$+rO4if7r8N@eisner.decus.org>   R In article <3A71DF0A.AC965CFC@virgin.net>, Alan Greig <a.greig@virgin.net> writes: >  >  > Rob Young wrote: > U >> In article <3A71D783.8CE9DDE6@virgin.net>, Alan Greig <a.greig@virgin.net> writes:r >>A >>         The Ohio cluster isn't a supercomputer bid.  It is 142uF >>         CPUs. 71 - two processor Itanium boxes in a Myrinet.  ThereE >>         are dozens of much larger Beowulfs around the world.  What 2 >>         that link points to is marketing fluff. >> >> > R > Rob I don't really disagree with virtually anything you say technically. But youX > either don't or won't get it: "Marketing fluff" is almost all there is. It is the mostS > important single factor. Do you really not understand this or do you think we are-E > taken in? You don't have to convince us of the technical arguments.  >   8 	Well... I politely disagree.  The folks that decided to; 	pick Compaq as a solution for the $200 million 30 Teraflopt= 	System had more NDAs and visits for higher ups than probablys@ 	most folks see in a lifetime.  They picked the best technically" 	sound solution for *their* needs.  5 	But yes, I will agree that the great unwashed massesS9 	love marketing and golf outings and free shirts and freet8 	jackets and for the truly fortunate ... free vacations.   				Robs   ------------------------------  % Date: Fri, 26 Jan 2001 21:15:47 +0000p% From: Alan Greig <a.greig@virgin.net> $ Subject: Re: SGI x Itanium x Cluster* Message-ID: <3A71E903.C23809B5@virgin.net>   Rob Young wrote:  T > > Rob I don't really disagree with virtually anything you say technically. But youZ > > either don't or won't get it: "Marketing fluff" is almost all there is. It is the mostU > > important single factor. Do you really not understand this or do you think we are G > > taken in? You don't have to convince us of the technical arguments.  > >    > A >         Well... I politely disagree.  The folks that decided toeD >         pick Compaq as a solution for the $200 million 30 TeraflopF >         System had more NDAs and visits for higher ups than probablyI >         most folks see in a lifetime.  They picked the best technicallya+ >         sound solution for *their* needs.o >o  ( That's why I said "almost all there is".   > > >         But yes, I will agree that the great unwashed massesB >         love marketing and golf outings and free shirts and freeA >         jackets and for the truly fortunate ... free vacations.e > % >                                 Robi   --
 Alan Greig   ------------------------------    Date: 26 Jan 2001 12:51:13 -07001 From: nothome@spammers.are.scum (Malcolm Dunnett)B$ Subject: Re: SGI x Itanium x Cluster, Message-ID: <5r0SkG1b0PLf@malvm1.mala.bc.ca>  , In article <V$+rO4if7r8N@eisner.decus.org>, 1      young_r@eisner.decus.org (Rob Young) writes:r  S >> Rob I don't really disagree with virtually anything you say technically. But yougY >> either don't or won't get it: "Marketing fluff" is almost all there is. It is the mostuT >> important single factor. Do you really not understand this or do you think we areF >> taken in? You don't have to convince us of the technical arguments. >> = > : > 	Well... I politely disagree.  The folks that decided to= > 	pick Compaq as a solution for the $200 million 30 TeraflopO? > 	System had more NDAs and visits for higher ups than probablyM  > 	most folks see in a lifetime.  =    Given that us "peasant" customers couldn't get an NDA or ae= visit from a higher up if our lives depended on it that's noto saying much, is it? :-)    ------------------------------  % Date: Fri, 26 Jan 2001 13:37:27 +0100 , From: "Nico van der Boom" <njvdboom@caiw.nl>< Subject: Re: Simple DECnet-IV Tasks Impossible under Phase-V- Message-ID: <94rqld$1uie$1@news.kabelfoon.nl>g  = "arturo saavedra" <arturo.saavedra@wcom.com> wrote in messagee= news:MOEAJKGGEIMGCCPEPJBHOENKDDAA.arturo.saavedra@wcom.com...8@ > You can also try setting the node address through command line >sK > $ mcr decnet_register register node local:.nodename address xx.xx synonymc
 > nodename >OC > where the synonym is basically the phase iv representation of thet > nodename/address >n > Art  >   & i use the mentioned command regularly.   some extra info.   example node FOOBAR address 1.1t  ! Register node in local namespace:LI $ mcr decnet_register register node LOCAL:.FOOBAR syn FOOBAR towers {1.1}n    (watch the dot in LOCAL:.FOOBAR)   delete node from namespace, $ mcr decnet_register deregister node FOOBAR  ( NOTE : the node FOOBAR is still in cache  A flush the cache (you need this when you renumber a decnet node !)X  6 $ mcr ncl flush session control naming cache entry "*"  L TIP: register the decnet node before adding proxies for that node. (it won't work otherwise)i when is doesn't work:t     - deregister the noder     - remove proxy     - register the nodec     - add proxye    L in SYS$SYSTEM there is an image something like *CDI*DUMP*.EXE (i am not on a system right now)y you can run this image. (mcr) C every decnet name lookup can be monitored on your screen (handy fora
 debugging)    H the export function of DECNET_REGISTER will create an text file with the node registrations.l> Use your editor or a script to populate this simple text file.  * then import your file and flush the cache.   --   CU,l Nico van der Boomv >===> http://www.vanderboom.com ( >===> http://home.kabelfoon.nl/~njvdboom  + after 8 years of NCL you will like it too !o   ------------------------------  % Date: Fri, 26 Jan 2001 04:11:29 -0500s2 From: rdeininger@mindspring.com (Robert Deininger)< Subject: Re: Simple DECnet-IV Tasks Impossible under Phase-VL Message-ID: <rdeininger-2601010411290001@user-2ivebf4.dialup.mindspring.com>  f In article <3A70E67C.378FCB33@earthlink.net>, "David J. Dachtera" <djesys.nospam@earthlink.net> wrote:  ; > "Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515" wrote:c > > 0 > > In article <3A70DBB5.3F1B4EC@earthlink.net>,A > >     "David J. Dachtera" <djesys.nospam@earthlink.net> writes:iG > > > Can *ANY*one give me the NCL equivalent of the following command?t > > >e > > > $ MC NCP$ > > > NCP> SET NODE 1.23 NAME NEWNOD > > 5 > >         Not NCL, but the documented method is to,I > > 2 > >                 RUN SYS$SYSTEM:DECNET_REGISTER > > L > >     (or use MCR for less typing  :-).   This  presents a menu which willL > >     allow  you to register a DECNET IV address in the "local" name spaceB > >     for DECNET-Plus.  ...Hope I got the terminaolgy correct... > I > Yeah - tried that. Couldn't figure out what the hell it asking about. IhD > just want to "register" my other DECnet (all Phase-IV, by the way)0 > nodes. Never saw an option like that in there. > I > Why all the "rocket science" that no one in our shop will ever use? All H > I want is to not have to use notation like 1065:: instead of MYNODE::,A > without having to take four years of classes to learn how to doh' > something as simple as NCP> SET NODE.     / This stuff IS documented, so calm down already.n   Introduction and User's Guide:?    http://www.openvms.compaq.com:8000/72final/6501/6501PRO.HTMLtK Read about the "local namespace".  That's usually enough for simple setups.t  % Installation and Basic Configuration:o?    http://www.openvms.compaq.com:8000/72final/6496/6496pro.HTMLf Look at chapter 7.  , IIRC, I did the IV --> plus conversion using%    $ @sys$manager:net$configure basic    as described in chapter 6.  It found the old node database and converted it.  I didn't really do anything.  I agree the manuals are weak in the area of conversion, but the net$configure thing did pretty well.  k I have notes saved somewhere, if you want more details.  I certainly don't remember off the top of my head.n   -- r Robert Deininger rdeininger@mindspring.come   ------------------------------  % Date: Fri, 26 Jan 2001 20:21:53 +0000t+ From: "antonio.carlini" <arcarlini@iee.org>i< Subject: Re: Simple DECnet-IV Tasks Impossible under Phase-V' Message-ID: <3A71DC61.B56BEC9F@iee.org>t   "David J. Dachtera" wrote: >  > G > The proc. asks a bazillion questions, but never *ONCE* mentions otheraH > DECnet nodes (though it does ask about MOP clients - don't have any of	 > those).- > ; > > Then do a PRODUCT DELETE (or is it remove ?) of decnet.e> > > It does a pretty good job. A couple of stray files remain. > H > I gotta find a way to do that - somehow make a copy of the system diskG > so I can bring that home and mess with it on my hobbyist system here.p >  > Messing with this is a crock!c    4 As others have pointed out, fire up DECNET_REGISTER.   Once you get that far you need:n  # "2 - Register or modify node names"w  ) Assuming you are using a local name spacee& (and it would be hard for you to have ' set  up DECdns without actually knowing   a fair bit about DECnet Phase 5) then pick LocalFile and give it the nodename LOCAL:.FOO" and synonym FOO, address area.node and transport NSP. That should get you enough to talk to   Phase IV nodes and Phase V nodes which have been configured to>! use NSP (the Phase IV transport).s    Alternatively you could just do  	$ @SYS$MANAGER:NET$CONFIGURE $ and let it convert your old Phase IV# database (assuming it's still lyingp around).   Antonios   -- ,   --------------- - Antonio Carlini             arcarlini@iee.orgS   ------------------------------  % Date: Fri, 26 Jan 2001 18:30:03 -0600e7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>h< Subject: Re: Simple DECnet-IV Tasks Impossible under Phase-V- Message-ID: <3A72168B.FC8C3128@earthlink.net>g   "antonio.carlini" wrote: >  > "David J. Dachtera" wrote: > >i > >bI > > The proc. asks a bazillion questions, but never *ONCE* mentions other-J > > DECnet nodes (though it does ask about MOP clients - don't have any of > > those).. > >D= > > > Then do a PRODUCT DELETE (or is it remove ?) of decnet.g@ > > > It does a pretty good job. A couple of stray files remain. > >1J > > I gotta find a way to do that - somehow make a copy of the system diskI > > so I can bring that home and mess with it on my hobbyist system here.s > >e! > > Messing with this is a crock!S > 6 > As others have pointed out, fire up DECNET_REGISTER. > ! > Once you get that far you need:n > % > "2 - Register or modify node names"c > + > Assuming you are using a local name spaceC' > (and it would be hard for you to havex) > set  up DECdns without actually knowinga" > a fair bit about DECnet Phase 5) > then pick LocalFile and give > it the nodename LOCAL:.FOO$ > and synonym FOO, address area.node  > and transport NSP. That should > get you enough to talk to-" > Phase IV nodes and Phase V nodes > which have been configured toe# > use NSP (the Phase IV transport).i > ! > Alternatively you could just doD& >         $ @SYS$MANAGER:NET$CONFIGURE& > and let it convert your old Phase IV% > database (assuming it's still lying 
 > around).  H Actually, this was installed as a clean system by people who didn't know4 better than to take all the defaults on the install.  H In the end, Arturo's suggestion accomplished in a few minutes what years; of poring over the documentation and HELP couldn't resolve.-   Amazing!  F I still don't know what the appropriate NCL would be, but I'm not dead yet...  F Suggestion: Arturo or someone else - put together and submit a "DECUS"E session on Phase-V for those who cut their VMS teeth on Phase-III and. -IV.   -- m David J. Dachtera. dba DJE Systems  http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/a  F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.y   ------------------------------  % Date: Fri, 26 Jan 2001 18:31:11 -0600o7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>r< Subject: Re: Simple DECnet-IV Tasks Impossible under Phase-V, Message-ID: <3A7216CF.E26E7EC@earthlink.net>   arturo saavedra wrote: > @ > You can also try setting the node address through command line > K > $ mcr decnet_register register node local:.nodename address xx.xx synonymr
 > nodename > C > where the synonym is basically the phase iv representation of thel > nodename/address   THAT WORKED MARVELOUSLY!!!  " THANK YOU!  THANK YOU!  THANK YOU!   -- r David J. Dachtera" dba DJE Systemsl http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/-  F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.e   ------------------------------  + Date: Fri, 26 Jan 2001 09:08:21 +0000 (UTC) ' From: Osmo Kujala <kujala@tukki.jyu.fi>a0 Subject: Re: Source for used StorageWorks disks?, Message-ID: <94req5$rou$1@mordred.cc.jyu.fi>  & Uwe Zessin <zessin@my-deja.com> wrote:  ; > Hm. '20,F10,W,W' I find on a DS-RZ28M-VZ (green) carrier.3' > Are you sure you have an U-SCSI disk?m  H Not yet in the the container. I was going to take away the 4GB wide ones7 and put in bigger ultra-wide ones, but maybe giving up.p  H > And I/O modules (enough power for the U-SCSI module - 180W are needed,E > if I recall correctly), appropriate cables and a suitable backplanep; > (the U-SCSI variant has additional shielding) in the box.s  0 Enough watts and cables could be changed but ...  G >> Does BA356-SB (and cable BN21K-02 and "connector piece" 70-31490-01)6 >> support ultra?  > Is that a grey box?b% Grey it is (or white and very dirty). 1 > Then I would not try this out. At least you arewE > violating FCC (or what ever they are called in your company) rules.d4 > Such a configuration might also corrupt your data. ...-C > Wide cable means 'nothing', sorry. DEC _did_ sell U-SCSI _narrow_o' > SBBs (e.g. DS-RZ1CF-VA, DS-RZ1DF-VA).m ...uC > Also, there are _three_ different U-SCSI I/O modules (also called1E > 'personality modules') for the BA356 U-SCSI box: 2 single-ended and B > a differential one that allows cable length of up to 20.5 meters > for a point-to-point segment.a  - What a mess. I'll give up. Thanx for warning.u   Osmo   ------------------------------  % Date: Fri, 26 Jan 2001 04:34:37 -0500 2 From: rdeininger@mindspring.com (Robert Deininger)0 Subject: Re: Source for used StorageWorks disks?L Message-ID: <rdeininger-2601010434370001@user-2ivebf4.dialup.mindspring.com>  [ In article <87elxr61f2.fsf@prep.synonet.com>, Paul Repacholi <prep@prep.synonet.com> wrote:     > > Compaq REALLY ought to get the cables catalogue back, and anA > SOC that contains enough info to enable peole to sort this sortc > of stuff out.a   Hey!  I stumbled on a copy of a cable summary on a web page a couple of weeks ago.  I saved the URL, but I can't get to it now because there's no power at work.  Remind me...   --   Robert Deininger rdeininger@mindspring.come   ------------------------------  # Date: Fri, 26 Jan 2001 15:06:51 GMT % From: Uwe Zessin <zessin@my-deja.com>-0 Subject: Re: Source for used StorageWorks disks?) Message-ID: <94s3q2$pkp$1@nnrp1.deja.com>@  , In article <94req5$rou$1@mordred.cc.jyu.fi>,*   Osmo Kujala <kujala@tukki.jyu.fi> wrote:( > Uwe Zessin <zessin@my-deja.com> wrote: >7= > > Hm. '20,F10,W,W' I find on a DS-RZ28M-VZ (green) carrier.:) > > Are you sure you have an U-SCSI disk?p >oE > Not yet in the the container. I was going to take away the 4GB wides> > ones and put in bigger ultra-wide ones, but maybe giving up. >sB > > And I/O modules (enough power for the U-SCSI module - 180W areE > > needed, if I recall correctly), appropriate cables and a suitable G > > backplane (the U-SCSI variant has additional shielding) in the box.u > 2 > Enough watts and cables could be changed but ... >eF > >> Does BA356-SB (and cable BN21K-02 and "connector piece" 70-31490- 01)  > >> support ultra?v > > Is that a grey box?a' > Grey it is (or white and very dirty).-  H I've talked to a colleague and he thinks that all white boxes are Ultra,H however he agrees with my comment to him that a 'white box' isn't called> a BA356, but rather something like 'PCgrglp' (dirty or not ;-)  3 > > Then I would not try this out. At least you are:G > > violating FCC (or what ever they are called in your company) rules..6 > > Such a configuration might also corrupt your data. > ...TE > > Wide cable means 'nothing', sorry. DEC _did_ sell U-SCSI _narrow_.) > > SBBs (e.g. DS-RZ1CF-VA, DS-RZ1DF-VA).n > ...eE > > Also, there are _three_ different U-SCSI I/O modules (also calledaG > > 'personality modules') for the BA356 U-SCSI box: 2 single-ended and D > > a differential one that allows cable length of up to 20.5 meters! > > for a point-to-point segment.  > / > What a mess. I'll give up. Thanx for warning.h >o > Osmo   --
 Uwe Zessin3 (If you want to send mail, please use user "zessin"k/ who lives at "decus.decus.de", not my-deja.com)u     Sent via Deja.comm http://www.deja.com/   ------------------------------  % Date: Sat, 27 Jan 2001 00:29:08 +1100n+ From: "dooley" <dooleys@cooma.snowy.net.au>a" Subject: Re: Telnet, yes; HTTP, no, Message-ID: <3a717b39.0@snowy2.snowy.net.au>  ; Carmine Castiglia <ccastiglia@engsint.com> wrote in messager% news:94pnvp$ff1$1@bob.news.rcn.net... ) > OpenVMS 7.1-1H2,  Decnet Phase IV (UCX)  > I > Connecting an HP JetDirect device to the network using TCP/IP services.e DidfH > the SET HOST thing in UCX, set the IP address in the JetDirect device. > L > I can telnet to the device, read the configuration table, etc.  I *cannot*= > reach the device's built-in web server - attempting to load.E > http://151.106.50.20 (as described in JetDirect docs) from Netscapel browsern9 > results in "unable to connect to server" type messages.e? Can you telnet to it on port 80? or maybe it uses another port?  Also (and this isfK > why we're here), I cannot make the device function as a server queue - itn? > seems to setup correctly but then attempts to print result in 
 %UCS-F-NOMSG,  > Message number 030AEBB4. >  <snip> this looks familiar $ if you run lprsetup to add a printer6 and you don't have the privilege to create directories& it doesn't create one for that printer (but doesn't tell you)4 have a look in your tcpip$printcap (or ucx$printcap) sd: is spool dir lf: is logfile* Phil   ------------------------------  # Date: Fri, 26 Jan 2001 15:56:50 GMT 4 From: LESLIE@209-16-45-102.insync.net (Jerry Leslie)2 Subject: Re: VMS 7.2-1 Bugcheck In DECW$REINIT.EXE& Message-ID: <67hc6.257$_O.7678@insync>   ethornber@my-deja.com wrote: : 1 : It appears you need need to adjust NPAGEDYN ...  :  : SOLUTION:  : * : From OpenVMS Alpha V7.1-2 Release Notes: : 4 : -------------------------------------------------- : --------------------0 : 2.19  Problem on Digital Personal Workstations : with Powerstorm Graphic 
 :       Cardsr4 : -------------------------------------------------- : -------------------- :  : 0 : To correct this problem, increase the NPAGEDYN : system parameter. Foru2 : example, if the system crashes with NPAGEDYN set : to 2244608, increase1 : the parameter setting to 4489216 to prevent the  : system from crashing.a/ : OpenVMS recommends that you run AUTOGEN afterd : increasing the NPAGEDYN  : system parameter.t :  : Ed.  :  :     L MIN_NPAGEDYN was raised from 7966720 to 63700992 in SYS$SYSTEM:MODPARAMS.DATM an AUTOGEN was performed, followed by a reboot, but the system still crashed.e  / Time for a call to the Customer Support Center.m    / --Jerry Leslie   leslie@209-16-45-97.insync.nete;                  leslie@209-16-45-102.insync.net is invalido2                  (my opinions are strictly my own)   ------------------------------  # Date: Fri, 26 Jan 2001 19:14:29 GMT 4 From: LESLIE@209-16-45-102.insync.net (Jerry Leslie)2 Subject: Re: VMS 7.2-1 Bugcheck In DECW$REINIT.EXE& Message-ID: <p0kc6.263$_O.7840@insync>  * Jerry Leslie (leslie@clio.rice.edu) wrote: : B : The following crash is from a client's system, running VMS 7.2-1   One new piece of data:  >  The system boots fine the first time after power is applied, 5  but crashes as described when a reboot is attempted.     / --Jerry Leslie   leslie@209-16-45-97.insync.nets;                  leslie@209-16-45-102.insync.net is invalidd2                  (my opinions are strictly my own)   ------------------------------  # Date: Fri, 26 Jan 2001 07:47:33 GMTi From: sabolich@my-deja.com@ Subject: Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken)) Message-ID: <94ra2k$4h8$1@nnrp1.deja.com>O  C > > have a look at http://uptime.netcraft.com/up/today/top.avg.htmld >OB > Checked that and was first impressed of the frequently mentionedB > FreeBSD. A second look raised some doubts about how useful these > figures are.= > What does this mean "For performance reasons, we limit thiseA > monitoring process to the most frequently requested sites." Who B > does request these sites? I personally doubt that the third siteB > on the list "Universitaetsklinikum Rudolf Virchow" is frequently > requested.? > If you take a closer look at the last column "Netblock Owner"a@ > you won't be very happy if you detect that the same owner e.g.B > "Hopemoon Internet" but others also show up several times. It is9 > in no way clear to me that the results are independent.gA > Technically the different URLs in the first column belonging toeA > the same "Netblock Owner" could be served with the same server.lA > But the worst: I don't see any relation with what I was talking/ > about.  H Good. At first you were a bit impressed but later correctly realized the# statistics mean absolutely nothing.,   >aC > > What is talked about and what is used are not allways the same.n >); > True but not helpful in our situation. I doubt that theseVA > netcraft numbers are reliable or useful in any respect. In facts< > I think that there are about 1'000'000 times as many LinuxA > system in usage than FreeBSD systems. What is the point of whati > you said?   H I think you are measuring Linux popularity by the amount of noise comingE from Linux.  And conclude that since 1 000 000 times less noise comesn5 from FreeBSD it must be 1 000 000 times less popular.l  F First of all, Linux is often peoples first exposure to a Unix (or UnixG like OS since technically did not descend from AT&T or BSD sources) - IiH am an example.  A lot of these people later try a *BSD as they gain someD experience (again like me).  As the BSD crowd generally has a littleF more experience they tend to cool off a bit and make less noise.  Many4 run both and split server/workstation -- *BSD/Linux.  H No doubt Linux is more popular but it is next to impossible to have good? statistics (any are as meaningless as the netcraft ones).  I'veDF downloaded Linux and FreeBSD -- burned them to CD, given them to otherG people.  Some CD may not have been used at all.  Some were installed onl multiple machines, etc.t  H Linux is more popular but there is (I think) more *BSD than you estimate. (like I can read your mind, but just a hunch).  B > You are completely right but it seems to me that you are missingB > my point. I expect a system which is optimised for a certain cpuA > type with respect to performance to be perfomant. Unfortunately @ > it is not performant when it comes to systems of that specific? > cpu type which are built for high performance requirements. Iw6 > would call that a clear example of missing the goal. >a  D You have a point.  Something like Solaris x86 should scale nicely toE multiple CPUs.  The only problem is Solaris x86 (from what people whoeH claim to have experience with it say) is so much slower than FreeBSD you> need a few CPUs just to keep up with a single CPU FreeBSD box.   You might want to check:  1 http://www.bsdtoday.com/2000/October/News296.htmld ands< http://www.geocities.com/SiliconValley/Bay/6986/yahoobsd.htm  E > > There are many implementations of Unix, varies by implementation.- >-A > How do they excel? The internal VMS architecture excels in manyi > respects.i >oB To honestly give you a first hand answer I would have to study theG sources.  Some are monolithic, some are microkernels, some are designed D for hard real time, etc.  I suppose you could find some good work inH many areas.  Just a guess though, cannot comment on code I haven't seen. > > 7 > > > 2.API - how the OS is presented to the programmert > >vE > > This was my main point.  Unix has a short, simple, effective API.s >m@ > Can't buy that. Short, simple, effective are prerequisites but@ > have to be measured against what they provide and this has not? > much to do with quality of design. As I pointed out later the @ > UNIX design goals were rather strange. Simplicity according toB > the tasks that have to be accomplished is a virtue. But loweringA > the goalpost instead of thinking about what almost every usefulo@ > program will need is a no-brainer. The UNIX people did exactly > that.D  C Ok, that's your opinion - fair enough.  My opinion is Unix has whatO= almost every useful program needs.  Can we agree to disagree?h   >  > >o5 > > > 3.CLI - how the OS is presented to the sysadmin  > >    snip  B > Obviously I wasn't clear at all, so you completely misunderstoodB > what I wanted to explain but it might not easy to understand for> > a person who is accustomed to UNIX. I claimed that different> > standard commands should have consistent options in that the: > same aspect should be controled with the same option forA > different command. An example with Linux system commands. ls -re? > shows the directory reverse sorted. ls -R shows the directory @ > and all its subdirectories and all their subdirectories and soB > on. cp -r and cp -R both copy one directory and all nested files? > and directories in there whereas the man page gives different   > descriptions for both options.  F Because there is a difference (usually not but there can be if copying? special files such as /dev/zero).  The man page is quite clear.   " > rm -r and rm -R do the same, but: > mv -r and mv -R don't exist (e.g. mv -R *.map /archiv/).   Correct, point taken.   A > ls, ps and escpecially find have a lot of different options anda; > different meaning of the "same" options on different UNIXnA > variants with ps and find beeing of the worse. Examples needed?  >N  B Well there are lots of Unixes and one VMS.  This is a double edgedG sword.  Some difference between vendors or single vendor lock-in.  Pickc your poison.  D > > I'll try to give a good example here.  Suppose you have two fileK > > descriptors (fds which is an open serial port, and fdf which is an opentJ > > file)  The program xyz takes input on standard in and writes it outputH > > to standard out.  You want to run xyz with fds for input and fdf for? > > output.  You also want to run it at boosted priority level.e > >rJ > > In VMS you will have one call to the OS the create process xyz.  ThereJ > > is a big structure you pass containing information about input, outputK > > handles (for redirection), priority, priviledge, etc.  Everything aboutt5 > > that process must have a place in that structure.h > >a* > > Unix is completely different, example: > >e+ > > x = fork() /* create copy of process */s > > if (x == 0)t > > {m > > /* we are child */ > > /* make fds standard in */ > > dup2(fds, 0);  > > /* make fdf standard out */  > > dup2(fdf, 1);." > > /* adjust my priority to -5 */% > > setpriority(PRIO_PROCESS, 0, -5); 0 > > /* replace current process image with xyz */ > > execl("xyz", NULL);. > > /* never get here */ > > } 
 > > else { > > /* we are parent */  > > }i > >AK > > So I've replaced a complex system call (in VMS) which takes a structureoD > > with a few simple ones in Unix (none of which take a structure). > >sK > > That is what I like about unix.  Simplicity and flexibility.  It's kindeL > > of like a lego or erector set.  The basic components are very simple butF > > you can put them together any way you want to achieve exactly what	 you want.a >0A > Honestly I don't think that having multiple calls is simpler oriB > safer than decorating a structure and executing only one call. I= > personally like structures much more but this is probably ae > matter of taste.  B Agreed, matter of taste.  In fact this whole which is 'cleaner' or# better 'designed' is just that too.a  0  But distributing calls which should be executed% > as a bundle is obviously less safe.a  @ Why? Giving a compiler to a programmer is like giving him a hand, grenade, might as well give him a gun too :)    But what is much more> > important: the VMS interface is much richer. The UNIX API isB > like a lego set for three year old children, whereas the VMS API7 > is like a Fisher Technik set for 15 year old teenies.   E Maybee, but that lego set will do everything the Fisher Technik does.oH Anyways, enough of this.  Some prefer simplicity while others features - I think we can agree on this.t   snip  H > > Parameter parsing functions are in the C library.  On the whole codeH > > reuse is good.  Suppose you want to see all unique processes running! > > excluding bash, output sortede >kA > Misunderstanding. I'm not talking about code reuse with respectn< > of parameter parsing. I'm talking of implementation effort< > because each program has to check its parameters by itself? > whereas with VMS this is done generically and the program haswB > only to use the set of supplied values. Did you ever implement a= > program under VMS which had to use command line parameters?d >   / Maybee years ago (can't really say I remember).p Short example please?a   > >k& > > ps -A | sort | unique | grep -v sh > >sK > > As you can see a few simple commands piped together.  ps (process show) L > > does not need to have sort logic programmed in, etc.  ps need not have aH > > million options trying to anticipate everything the user could want.% > > Connect them up however you want.A >e? > Sorry, but I completely disagree. First you can't combine allu> > the programms using pipes because there are problems how the@ > parameters are interpreted. E.g. if you pipe to grep the input= > is interpreted as a file but you can't pipe to the pattern..  A Yeah, because the pattern comes from the command line, not stdin.e  < > Another example? "find . -name '*.text' | grep foo" simply* > doesn't work. Why? More examples needed?  3 "find -name '*.conf' | grep httpd" works fine here.iF I spits some "Permission denied" message out, but those go to standard error.C "find -name '*.conf' 2>/dev/null | grep httpd" gets rid of standardr error junk.r    I have the impression? > that pipeing is a patch work approach which works most of theg> > time but not all the time and it's not well thought through.  F Ok, so it doesn't work _every_ time, but it allows you to come up with0 combinations unanticipated by the programmer(s).   snip  8 > Why does "ls" offer sorting options - couldn't that beB > done with pipeing and sort (should be srt!)? Why does "ls" offer> > output format selection selection options - couldn't that beB > done with pipeing and grep (should be grp!)? Wouldn't it be muchA > more efficient if the "ls" like the VMS DIRECTORY command couldrA > all sorts of sorting and selecting because it knows best how toa > interpret the data?:  G I am the author of the VMS DIRECTORY command.  I have anticipated every 0 possible output format any user could ever want.  F Common options are given by command line (or else commands would startC to look like sausage links).  Pipeing is useful for doing somethingm< beyond the command line options.  It gives more flexibility.   snip  B > While I completely agree with your conclusion I'm sure because I< > have facts and numbers that this is not sufficient. Like a< > typical C programmer who isn't able to do real programming; > because he or she is subconsciously absorbed by mastering8B > difficulties which are a waste of time to be mastered by a humanA > beeing and by that are unable to achieve a level of abstractioni; > and overview which is necessary to do real programming asgA > opposed to dumb coding, it is this attitude of putting togethero@ > simple but insufficient parts which creates this vast majority? > of unusable, unmaintainable, inefficient and architecturelesso@ > programs. This C programmer attitude is exactly what makes theB > UNIX programmers happy by re-inventing the wheel again and againB > instead of trying to reach a higher level of programming. Why is@ > parsing an unstructured text file a pleasure (UNIX approach to= > configuration files) while fetching a value from structured 6 > storage is a pain (VMS approach with RMS key files)?  F You can open a text file with any editor.  There are comments in theseE text files that explain the configuration.  When you make changes your. can add comments like when, why, by whom, etc.  D The programmer can create and comment the initial configuration fileG with a text editor.  There is no need to have some program ask the usereH for the configuration and then write it to the keys.  You either have to. write the parser or the configuration program.  H By the way, I'm getting the impression that I'm pissing some people off.D  This is not my intention and I'm sorry if I've done so.  I was just responding to you statement-  ! >Linux has the old, obsolete UNIX:= > architecture and is by that like all other UNIXes much lessj > modern than VMS.  = >Of course I'm on my way back to VMS and I will do as much asn? > possible on VMS because the architecture is clean in every ofv@ > the three important dimensions. And the UNIX architecture will4 > never be clean; if it will it is not UNIX anymore.  3 because I feel that Unix is the cleaner of the two.t  H I like VMS, I just like Linux more.  It was not my intention to slam VMS& and I'm sorry if it appeared that way.   No hard feelings.*   Fran             Sent via Deja.comv http://www.deja.com/   ------------------------------  # Date: Fri, 26 Jan 2001 09:46:47 GMTl From: sabolich@my-deja.com@ Subject: Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken)) Message-ID: <94rh26$ah6$1@nnrp1.deja.com>   @ > My news server seems to have discarded this, so I'm reposting.I Slightly out of time order, but still replying to the right post, I hope.t >rD > > I'll try to give a good example here.  Suppose you have two fileK > > descriptors (fds which is an open serial port, and fdf which is an openaJ > > file)  The program xyz takes input on standard in and writes it outputH > > to standard out.  You want to run xyz with fds for input and fdf for? > > output.  You also want to run it at boosted priority level.- >- > Ok.- > J > > In VMS you will have one call to the OS the create process xyz.  ThereJ > > is a big structure you pass containing information about input, outputK > > handles (for redirection), priority, priviledge, etc.  Everything aboutp5 > > that process must have a place in that structure.I > @ > Well, no.  I guess you are describing SYS$CREPRC, which is theH kitchen-sink method of creating a process.  I count 13 arguments to thisC routine, which sure is complex.  Only two of them are "structures", E IIRC. But they are ALL optional.  (Perhaps not all at the same time.) & So you are overstating the complexity.  G I'm not saying that SYS$CREPRC is hard, or complex, to use.  I'm saying,F that the SYS$CREPRC is complex because it has the ability to take many (13) arguments.  >4F > To compare the API complexity to create a process, you ought to noteC that a VMS process has a lot more to it than a unix one.  The extramH features have to be specified (or defaulted) somehow.  VMS supports veryH fine granularity in both privleges and quotas, for example.  Your littleC example doesn't use such features, so we'll let VMS do it's defaults thing.  That's easy. > F > But this is the wrong way to make this process, probably.  LIB$SPAWNH might be more suitable.  Same number of arguments, all of them optional, none of them a structure.- >-E > One specifies input, one specifies output, one specifies command toLD execute.  Some others help with interprocess communication, which we don't seem to need here. >sE > Since we don't clone processes in VMS, the new process will need touH set it's own priority, if we use lib$spawn.  sys$creprc lets the creator specify priority.i  C This is exactly why I put in the priority change -- to force use ofpF sys$creprc.  Also let's say we're not going to modify xyz so it cannot set its own priority.-   >-L > Depending on the context, we might not need another process at all in VMS. > * > > Unix is completely different, example: > > + > > x = fork() /* create copy of process */2 > > if (x == 0)3 > > {  > > /* we are child */ > > /* make fds standard in */ > > dup2(fds, 0);e > > /* make fdf standard out */r > > dup2(fdf, 1);a" > > /* adjust my priority to -5 */% > > setpriority(PRIO_PROCESS, 0, -5);m0 > > /* replace current process image with xyz */ > > execl("xyz", NULL);  > > /* never get here */ > > }o
 > > else { > > /* we are parent */i > > }r > K > > So I've replaced a complex system call (in VMS) which takes a structurepD > > with a few simple ones in Unix (none of which take a structure). >eD > So we might do something like this in VMS (Fortran, just to remind* folks that VMS is very language-flexible): > - >       status = lib$spawn('xyz','fds','fdf')l > G > We'll have to move the priority-change into xyz if we do it this way,eF or wrap xyz in a little script to change the priority if we don't want to change xyz.  A For my example to work I need sys$creprc and no wrapping xyz in al# script.  So we have something like:u  9 status = sys$creprc('xyz','fds','fdf',,,,,priority,,,,,,)   / Now suppose were not happy with the priviledge:e  C status = sys$creprc('xyz','fds','fdf',,,,,priority,,,priviledge,,,)w  - and for Unix we add a setuid() before execl()t   >tG > I don't think your example is simpler than mine.  Yours has comments;a= that's nice. Maybe I should have done mine in Ada, with namedl/ parameters.  Then I wouldn't need any comments.t >m  D Agreed, I don't think that my unix _example_ is simpler than the VMSE one.  What I'm saying is that VMS in one complex (but not necessarilyB> complicated from the programers perspective) call -sys$creprc-@ accomplishes the same thing as Unix in a number of simple calls.H Basically for every parameter you add inbetween the commas in sys$creprc< you will have to add a simple 1 or 2 parameter call in Unix.  G I think some confusion arose before because of semantics -- probably myi% fault because I was not clear enough.&  0 simple function - function taking few parametersD complex function - function taking many parameters (not neccessarily requiring all of them)  F The two examples are very similar in what they do, but quite differentK in how they do it.  I think that both _examples_ about equally complicated.oG  The Unix example I can do without documentation because it is easy fortE me to remember a series of simple, or low parameter count, functions.d@ The VMS example is also not a problem - i doubt I could rememberD everything and would have to look at the documentation for parameterD positions and counting commas.  But it is no big deal one way or the@ other.  I just think that the Unix way is more elegant, cleaner.  G This is why I say Unix has a good design.  These simple building blockseF put together in various ways can do just about everything a programmer could want.w  H Also for the user; using pipes to connect up various programs to do what	 he wants.   D Of course this is my opinion and I know yours is different.  I do toK some extent understand your opinion.  I hope you understand mine a bit too.t   snip   Fran     Sent via Deja.comt http://www.deja.com/   ------------------------------    Date: 26 Jan 2001 16:10:21 +0100G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>h@ Subject: Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken)H Message-ID: <y4puha2vtu.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>   sabolich@my-deja.com writes:  B > The VMS example is also not a problem - i doubt I could rememberF > everything and would have to look at the documentation for parameter  > positions and counting commas.  H That is what LSE is for - remembering the call prototype, and serving it
 up on demand.n   	Jan   ------------------------------  % Date: Fri, 26 Jan 2001 13:43:28 -0500s2 From: rdeininger@mindspring.com (Robert Deininger)@ Subject: Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken)L Message-ID: <rdeininger-2601011343280001@user-2ive6bd.dialup.mindspring.com>  E In article <94rh26$ah6$1@nnrp1.deja.com>, sabolich@my-deja.com wrote:      E > This is exactly why I put in the priority change -- to force use offH > sys$creprc.  Also let's say we're not going to modify xyz so it cannot > set its own priority.u   So you deliberately picked an example that demonstrates your point.  But your statements are more general: Unix is "cleaner" than VMS.  Of course we could pick examples that pointed in the other direction.  That is a danger of reasoning by example.  / > >       status = lib$spawn('xyz','fds','fdf')e > > I > > We'll have to move the priority-change into xyz if we do it this way,oH > or wrap xyz in a little script to change the priority if we don't want > to change xyz. > C > For my example to work I need sys$creprc and no wrapping xyz in a 	 > script.b  Perhaps not fair.  A script is certainly a way to combine simple things to achieve a desired complex whole.  A moment ago, that was a Good Thing.  Why would you not allow a script, if it happens to be the right VMS tool?  Are you trying to start an argument? :-)   > So we have something like: > ; > status = sys$creprc('xyz','fds','fdf',,,,,priority,,,,,,)s > 1 > Now suppose were not happy with the priviledge:o > E > status = sys$creprc('xyz','fds','fdf',,,,,priority,,,priviledge,,,)r > / > and for Unix we add a setuid() before execl()p  5This starts to raise another issue.  Neither of us was careful to check for possible errors.  Everything that goes wrong is encoded in my STATUS, but I did not look inside for every possible error.  You didn't look for errors either, but I assume each of those simple calls you used can return various errors.t   Among all the sys$creprc arguments, there are some that interact -- priviledges, file protections, priorities, etc.  The monolithic VMS call will check everything, and then create the process or not.  If it doesn't it will tell you why.  With the piecemeal unix approach, you may get partway into the job and find that something goes wrong.  So you have a partly-done job, and deciding what to do and cleaning up might be harder.  You need to check a lot of separate status values to catch all the errors.  It de   To me, each call in a well-designed API does ONE thing well.  sys$creprc creates one process to the caller's specifications.  That seems to fit my idea of ONE thing pretty well.  The fine granularity of the unix approach to the problem makes you build your one thing out of many small ones.  That's fine if the little things meet your personal definition of "one thing".  Once you get more than a handful of these little pieces, it seems harder to manage the collection UNLESS all the pieces are really independe    I > This is why I say Unix has a good design.  These simple building blocksaH > put together in various ways can do just about everything a programmer
 > could want.e  Well, we could decide to call the arguments to sys$creprc "simple building blocks", and the same statement applies.  Are 13 arguments more complex than 13 individual calls?  When we can omit any of them we don't need just by typing a comma?  I guess it comes down to how much they interact with each other.  If they don't interact, it's probably silly to collect them in one big call.  P The more I look at this example, the more similar the two approaches seem to me.   There is perhaps a gap in the VMS API.  A SYS$SETJPI - paralleling SYS$GETJPI - would occasionally be useful.  The functionality exists in the SET PROCESS command, but there are some tweeks that don't have a programmer's interface.b  F > Of course this is my opinion and I know yours is different.  I do toM > some extent understand your opinion.  I hope you understand mine a bit too.r  M Yes, a bit.  Of course the thing that is familiar usually seems the simplest.n   -- k Robert Deininger rdeininger@mindspring.come   ------------------------------  # Date: Fri, 26 Jan 2001 23:52:35 GMTh From: sabolich@my-deja.com@ Subject: Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken)) Message-ID: <94t2k0$n9s$1@nnrp1.deja.com>s   >e1 > > >       status = lib$spawn('xyz','fds','fdf')e > > > K > > > We'll have to move the priority-change into xyz if we do it this way, J > > or wrap xyz in a little script to change the priority if we don't want > > to change xyz. > >rE > > For my example to work I need sys$creprc and no wrapping xyz in a1 > > script.e >sB > Perhaps not fair.  A script is certainly a way to combine simpleE >things to achieve a desired complex whole.  A moment ago, that was a G >Good Thing.  Why would you not allow a script, if it happens to be then: >right VMS tool?  Are you trying to start an argument? :-)  H No argument wanted.  I think you misunderstood my intent.  I intended toH pick and example that lends itself to sys$creprc.  You may find a way toE skirt around sys$creprc with lib$spawn.  But there are times when youe0 have to use sys$creprc -- that is why it exists.  B My intent is to show that I can construct a complex (high argumentG count) sys$whatever function with no Unix equivalent, with a collection @ of simple Unix functions.  In the end the complexity of the Unix= contruction is comparable to its VMS sys$whatever equivalent.m   >p > > So we have something like: > > = > > status = sys$creprc('xyz','fds','fdf',,,,,priority,,,,,,)f > >m3 > > Now suppose were not happy with the priviledge:  > >nG > > status = sys$creprc('xyz','fds','fdf',,,,,priority,,,priviledge,,,)i > > 1 > > and for Unix we add a setuid() before execl()u >wC > This starts to raise another issue.  Neither of us was careful tolH >check for possible errors.  Everything that goes wrong is encoded in myA >STATUS, but I did not look inside for every possible error.  You.G >didn't look for errors either, but I assume each of those simple callst$ >you used can return various errors.  G Yeah, neither of us checked for errors.  But this also is pretty much a?D wash.  For example in sys$creprc if you try to create a process withD greater priviledge than you have it will return with error=you don'tH have priviledge to do this.  For every extra argument you decide to pass- you may get another possible error in status.e  ; setuid() can return success or error=EPERM (no permission).r  H So if you look at both methods (VMS and Unix) the sum of possible errors is pretty much the same.   >tE > Among all the sys$creprc arguments, there are some that interact --tH priviledges, file protections, priorities, etc.  The monolithic VMS callA will check everything, and then create the process or not.  If ittH doesn't it will tell you why.  With the piecemeal unix approach, you mayD get partway into the job and find that something goes wrong.  So youH have a partly-done job, and deciding what to do and cleaning up might be harder.   E Well in this case for Unix exit() after any failing call would be OK.dB You could also accept failure to set the priority, maybee log some error, and execl() anyway.  D In VMS it might fail with error=no priviledge to raise priority.  InB which case you coul re-issue the call without specifying priority.  ( Again I believe it is pretty much a wash  E >  You need to check a lot of separate status values to catch all thedF >errors.  It depends on the context which might be more work in robust >code.  G Yes, and the action you take is related to what function fails.  In VMSsJ the action is related to how (what error code) the single function failed.   >r> > To me, each call in a well-designed API does ONE thing well.D sys$creprc creates one process to the caller's specifications.  ThatG seems to fit my idea of ONE thing pretty well.  The fine granularity of F the unix approach to the problem makes you build your one thing out ofE many small ones.  That's fine if the little things meet your personalsE definition of "one thing".  Once you get more than a handful of thesetF little pieces, it seems harder to manage the collection UNLESS all the= pieces are really independent.  But perhaps they aren't quitebF independent, since they are all involved in making and configuring the same new process.o  ( I think they are independent. See below.   >fK > > This is why I say Unix has a good design.  These simple building blocksuJ > > put together in various ways can do just about everything a programmer > > could want.t > C > Well, we could decide to call the arguments to sys$creprc "simplexH building blocks", and the same statement applies.  Are 13 arguments more! complex than 13 individual calls?r   No, about the same.   H >  When we can omit any of them we don't need just by typing a comma?  IC >guess it comes down to how much they interact with each other.  IfwK >they don't interact, it's probably silly to collect them in one big >call.o >rF > The more I look at this example, the more similar the two approaches seem to me.T  C Yes.  You could say that.  There is one distinction though.  In thepF example the 13 arguments are analagous to 13 Unix functions.  However,F sys$creprc has one purpose - creating a process.  You would not use it> to set the priority of the current process like you could withC setpriority().  In this way sys$creprc is designed to do one thing.vG setpriority() also does exactly one thing, but the final result dependseA upon the context in which it is used - so it is a building block.a  F sys$creprc is also a building block in the context that it may be partF of an application.  It just that Unix takes this 'building block' idea down to a lower level.  7 I.e. - you can build a wall with small or large bricks..  E It is common in these VMS vs Unix exchanges for the VMS guy to rattle H off VMS's features including its rich comprehensive API which provides aF rich and comprehensive set of features - and perhaps the Unix guy does> not have enough understanding of VMS to fully appreciate this.  G The Unix guy then responds with 'You don't really need such a large settG of API functions).  The VMS guy might misinterpret this as meaning 'youeD don't really need all thos features'; which it does not really mean.G What the Unix guy means is that he can construct using simple functionssG the equivalent to any VMS function - with that construction and the VMSaH API call being of similar complexity.  Therefore Unix also has all those	 features.k  H Personally I like the Unix way better (I didn't use to though).  HoweverG I see how some people like the VMS way better.  I hope you can see whattE I'm trying to explain and can understand why some people prefer Unix.e  > What is cleaner or better designed is very difficult (probably@ impossible) to proove.  I stated that Unix is cleaner and betterG designed than VMS because Unix is more to my taste, but that is just mys opinion.  E If you asked me which one was cleaner and better a couple years ago IC would have said VMS.   snip  H > > Of course this is my opinion and I know yours is different.  I do toF > > some extent understand your opinion.  I hope you understand mine a bit too. >eE > Yes, a bit.  Of course the thing that is familiar usually seems theg	 simplest.4 >  Agreed.e   Fran     Sent via Deja.com  http://www.deja.com/   ------------------------------  % Date: Sat, 27 Jan 2001 01:07:55 +0000 ) From: Christof Brass <brass@infopuls.com>)@ Subject: Re: VMS vs UNIX (was: Linux worm and RedHat 7.0 broken), Message-ID: <3A721F6B.8140EC17@infopuls.com>   sabolich@my-deja.com wrote:eJ > Good. At first you were a bit impressed but later correctly realized the% > statistics mean absolutely nothing.   < Not good. You recommended something I wasted time on without@ improving your position. Why did you recommend that page anyway?   J > I think you are measuring Linux popularity by the amount of noise comingG > from Linux.  And conclude that since 1 000 000 times less noise comes.7 > from FreeBSD it must be 1 000 000 times less popular.o  = No, I don't. I observe the money put in by big companies, thes7 activities of big companies, the many software packages @ available for and developed under and the technical discussions. This is not noise.   H > First of all, Linux is often peoples first exposure to a Unix (or UnixI > like OS since technically did not descend from AT&T or BSD sources) - I.J > am an example.  A lot of these people later try a *BSD as they gain someF > experience (again like me).  As the BSD crowd generally has a littleH > more experience they tend to cool off a bit and make less noise.  Many6 > run both and split server/workstation -- *BSD/Linux.  < Agreed. My friend, to whom I referred previously, is another$ example. And I'm half of an example.   J > No doubt Linux is more popular but it is next to impossible to have goodA > statistics (any are as meaningless as the netcraft ones).  I'vesH > downloaded Linux and FreeBSD -- burned them to CD, given them to otherI > people.  Some CD may not have been used at all.  Some were installed on  > multiple machines, etc.t   True, but see above.  F > You have a point.  Something like Solaris x86 should scale nicely toG > multiple CPUs.  The only problem is Solaris x86 (from what people who J > claim to have experience with it say) is so much slower than FreeBSD you@ > need a few CPUs just to keep up with a single CPU FreeBSD box.  @ It's strange. I know a company which moved from Solaris/SPARC to8 Solaris/x86 because it was about equal as fast with same= processor speed and the boxes are way cheaper. I know another=> company who did tests with WNT and Solaris/x86 on the same box< stating that Solaris was about twice as fast. I don't know a? direct comparison of those two you mentioned but I really doubt > that. And Solaris scales very well with multiple CPUs - I know@ of tests from an independent company who switched from WNT boxes@ to Solaris/SPARC because of scalability with respect to multiple CPUs.m   >  > You might want to check: > 3 > http://www.bsdtoday.com/2000/October/News296.htmle  , I checked that and was twofold disappointed.? First: the page loaded very slowly - in fact it didn't finish -r@ and kept my Internet connection open for about one hour before I& detected it and stopped that nonsense.6 Second: while the number was impressing though no real; statistics about this exists as the author admitted it is a > little bit stupid not to check how many requests were directed6 to the Linux distribution and how many to the FreeBSD.: And, honestly, a claim of beeing up for two months without? crash, is a joke for this NG. Even Solaris can do that which isi8 like WNT or the IBM OSs dependent of periodially reboot.   > andU> > http://www.geocities.com/SiliconValley/Bay/6986/yahoobsd.htm  ? This *is* really a nice story. Some complaints nevertheless. Itt= is proposed like a finished archive reference article but the > author is not sure how to write sourcecode/source code (a wide8 spread contradiction in itself anyway) and uses the word> "stripping" when I assume he actually means "stiping" which is9 used within the same context a few sentences later. If ita; weren't a high quality implying article I hadn't pointed on > that. There is no time stamp but relative time statements like@ for the last two years. The list of tried out OSs is quite short: so the value of this comparison is very limited. While the; statements about their experience with the organisation andd? installation, stability and responsiveness of the user comunity > match what I've heard from my friend, the author is not really= independent (at least anymore) and therefore I'm not sure how ! much I can trust his assessments.i? What is very interesting wrt to our sub-subject, spreadness (isn< this an understanable word?): the author complains about the> small number of apps available and the neglection of the OS by@ almost all SW and HW companies. Isn't that a clear indication of< FreeBSD much less in use than Linux? What might also be very@ interesting to this NG is the author's statement that FreeBSD on< Alpha seems very interesting in price/performance rate. With@ open source I see the problems of an alternative CPU design like@ Alpha not as big as some years before. Alas without a time stamp> the article is not as helpful or meaningful as it could be. Do& you know when this has been published?   > G > > > There are many implementations of Unix, varies by implementation.e > >hC > > How do they excel? The internal VMS architecture excels in manyi
 > > respects.  > >nD > To honestly give you a first hand answer I would have to study theI > sources.  Some are monolithic, some are microkernels, some are designedoF > for hard real time, etc.  I suppose you could find some good work inJ > many areas.  Just a guess though, cannot comment on code I haven't seen.  @ This is an answer which speaks in favour of your credibility but@ not for the quality of the documentation about design and design: decisions. Unless there is a book like "Internals and Data@ Structures" it is hard to discuss the quality of UNIX design. Do: you know of any substantial documentation of that level of5 abstraction, i.e. instead of reading the source text?d  1 Which UNIX kernel is designed for hard real time?r  B > > Can't buy that. Short, simple, effective are prerequisites butB > > have to be measured against what they provide and this has notA > > much to do with quality of design. As I pointed out later the B > > UNIX design goals were rather strange. Simplicity according toD > > the tasks that have to be accomplished is a virtue. But loweringC > > the goalpost instead of thinking about what almost every usefulsB > > program will need is a no-brainer. The UNIX people did exactly	 > > that.n > E > Ok, that's your opinion - fair enough.  My opinion is Unix has what ? > almost every useful program needs.  Can we agree to disagree?i  : I agree that UNIX offers the basics i.e. what almost every@ useful progam needs. But I critise that the way it is offered is= not structured or designed or whatever enough and, maybe more 5 important, that it offers *much* to less to be really ? productive. You may get easier to speed in coding like on M$DOSi@ compared to MacOS but you get much less and you will never reach5 the overall productivity you reach with VMS or MacOS.t? Proposal: We define a small size project (not a tiny one like aa> few lines of C) and analyse where one or the other OS excells.= What I want to know if there is unnecessary complexity in thet@ VMS API as compared to the/one UNIX API if one want's to do real: work (not a "Hello World" like program). This is the first? litmus test. The second one is a full blown medium size projectt= (a big project is then only a scaling problem not a technicalE@ issue, because a full blown medium size project will have to use= all system services and facilities offered by the OS and evene= more, i.e. the limitations of each OS will show up when thesei@ missing parts have to be created; the point is to see how the OS2 can be used to write substantial real world apps).> I have to add something: my position is not simple my opinion.@ It is my opinion but not only. My position is based on facts, on? criteria and on experiences. Your criterion is a little bit toor1 less to measure the quality of an OS design with.h   >  > >  > > >c7 > > > > 3.CLI - how the OS is presented to the sysadminp > > >A >  > snip > H > Because there is a difference (usually not but there can be if copyingA > special files such as /dev/zero).  The man page is quite clear.   7 Sorry, it wasn't clear to me. Could you please explain?s  $ > > rm -r and rm -R do the same, but< > > mv -r and mv -R don't exist (e.g. mv -R *.map /archiv/). >  > Correct, point taken.. > C > > ls, ps and escpecially find have a lot of different options andS= > > different meaning of the "same" options on different UNIXtC > > variants with ps and find beeing of the worse. Examples needed?v > >r > D > Well there are lots of Unixes and one VMS.  This is a double edgedI > sword.  Some difference between vendors or single vendor lock-in.  Picku > your poison.  > I don't agree with the argument because it is mixing technical> and commercial aspects. I agree that single vendor lock-in has= its disadvantages as we all clearly see if this vendor is not,> doing what all conscious and not completely dumb people expect= it(?) (the vendor) to do. But there are three ways out: I wasg< talking about *design*. Why isn't there such a thing like UI< design (admitted on a very low level like the CLI) for UNIX?@ Conforming to one such standard doesn't prevent any competition.= And is POSIX not exactly trying that to achieve? I'm positivea@ that we lost more money by adopting scripts to different shells,@ by making mistakes because of different shells we've learned, by= learning different shells and by developping different shells = than has saved by the competition and the improvement through = it. Look at the very minor improvement of the shells. PERL iso< another hack which is a result of the constant weeknesses of= different shells. I'm sure PERL is not as close as popular one VMS as on UNIX and on WNT.  > Do you know the Russian M$DOS clone? It came too late but were? much better than the original. Same design. I heard rumors of a > Russion WNT clone. Besides legal issues there will be the same@ design but different implementations. This to say for seperating  commercial and technical points.  C > > Honestly I don't think that having multiple calls is simpler ornD > > safer than decorating a structure and executing only one call. I? > > personally like structures much more but this is probably a  > > matter of taste. > D > Agreed, matter of taste.  In fact this whole which is 'cleaner' or% > better 'designed' is just that too.   ? No, it isn't. I'm sorry that I'm that unclear not to be able toi? show the point. This one example and my attitude is a matter of > taste. But I'm sure you won't claim that the whole UNIX design@ is reflected within these few lines. And as I emphasised before:> "Hello World" is not a sufficient rich test case. And it would< be very helpful to have criteria to measure the quality of a design.o  3 > > But distributing calls which should be executed ' > > as a bundle is obviously less safe.  > B > Why? Giving a compiler to a programmer is like giving him a hand. > grenade, might as well give him a gun too :)  @ "Why?" Because it is more error prone, both from the perspective5 of the OS implementor and from the perspective of thea: programmer. For implementing validity checks the OS has to@ maintain state between the calls. The programmer must not forget= one of these calls. One could argue that the programmer coulds> also forget to properly initialise the structure. But there is< one important difference. The structure is a single point of? failure which is what we like as programmers which is obviouslye? opposite to what sysadmins like. A sequence of calls needs moret< attention and is not as easy to understand (when reading the@ documentation or reading the program text) and to explain in the documentation.  < "Giving ..." Completely agreed. But the PL "C" (if one might> call it so) is not well engineered let alone cleanly designed.@ It has *not* been engineered and has *not* been designed either.@ It has an unfortunate common fate with UNIX in its roots (how it@ began) and in its quality. I know the risk to publicly speak out> that "C" is completely crap but this is common sense among the= well educated engineers and among the PL researchers. I don'tl> know BLISS and I suspect it is even more low level than C, but@ it seems that is safer in the hands of VMS engineering than C in? the hands of UNIX programmers. C is like a hand grenade withoutI= timer. It will blow you off sooner or later if you try to usel< it. C is like a chainsaw without protection mechanism. If it8 were a normal consumer product the creators, vendors and@ suppliers of C compilers were all in jail for ever (according to> US American habbits) or executed if located for e.g. in Texas.   >  > > But what is much morey@ > > important: the VMS interface is much richer. The UNIX API isD > > like a lego set for three year old children, whereas the VMS API9 > > is like a Fisher Technik set for 15 year old teenies.r > G > Maybee, but that lego set will do everything the Fisher Technik does. J > Anyways, enough of this.  Some prefer simplicity while others features - > I think we can agree on this.p  ? Let's see if it can with the same efficiency wrt to developments; time, source quality, upgrade maintainability and executionb efficiency.u  J > > > Parameter parsing functions are in the C library.  On the whole codeJ > > > reuse is good.  Suppose you want to see all unique processes running# > > > excluding bash, output sortede > >eC > > Misunderstanding. I'm not talking about code reuse with respects> > > of parameter parsing. I'm talking of implementation effort> > > because each program has to check its parameters by itselfA > > whereas with VMS this is done generically and the program has D > > only to use the set of supplied values. Did you ever implement a? > > program under VMS which had to use command line parameters?m > >e > 1 > Maybee years ago (can't really say I remember).o > Short example please?l  < Do you need a program snippet? Otherwise I explain it short:@ Each C program in UNIX I've seen the source analyses the command; line with a switch statement where the parameter values areo; checked for validity. With VMS you don't need this validityt; check because the program will only run if the command lineb? parameters make sense. This checking mechanism is very flexibled@ and is located in VMS. You have to define a CLD file and to call= a command to tell VMS that it has to understand a new commandj7 and which parameters are allowed and which values these  parameters can take.   > > >b( > > > ps -A | sort | unique | grep -v sh > > > M > > > As you can see a few simple commands piped together.  ps (process show)tN > > > does not need to have sort logic programmed in, etc.  ps need not have aJ > > > million options trying to anticipate everything the user could want.' > > > Connect them up however you want.  > >nA > > Sorry, but I completely disagree. First you can't combine all @ > > the programms using pipes because there are problems how theB > > parameters are interpreted. E.g. if you pipe to grep the input? > > is interpreted as a file but you can't pipe to the pattern.s > C > Yeah, because the pattern comes from the command line, not stdin.t  < This is true but not a real response to my complain. Why the= hell is it the list of files which is read from stdin? And my @ point is: this concept is patchwork because it first creates the> illusion of powerful generality and when it comes to real life8 usage it miserably fails by only allowing a fixed set of9 combinations of very limited use and very low efficiency.i  > > > Another example? "find . -name '*.text' | grep foo" simply, > > doesn't work. Why? More examples needed? > 5 > "find -name '*.conf' | grep httpd" works fine here.yH > I spits some "Permission denied" message out, but those go to standard > error.E > "find -name '*.conf' 2>/dev/null | grep httpd" gets rid of standardf
 > error junk.o  < The error messages are no problem; we can avoid them by only@ 'finding' files which are allowed to read by grep I suppose. But> pipeing the result into grep simply doesn't work on my machine> (no error message, no message at all, nothing). And there is a  -exec switch for find. For what?= Sorry, I think, I found out, why it didn't work. Is it reallym= doing what it is expect to do in your example? I suspect thato? grep is switching its behaviour if it is fed from a pipe. If it-@ is fed from the command line the last parameters are interpreted@ as file names. If it is fed from a pipe it doesn't interpret any9 pameter as a file. Instead it assumes that the text to bet= searched in. Please check if your "working" example is reallya= doing what it should, i.e. searching in all the files for the ? occurence of httpd. If my analysis is correct then I think this0  is a good example for patchwork.   >  > > I have the impression A > > that pipeing is a patch work approach which works most of theb@ > > time but not all the time and it's not well thought through. > H > Ok, so it doesn't work _every_ time, but it allows you to come up with2 > combinations unanticipated by the programmer(s).  > True, but not helpful. All general concepts allow this. And in= fact it weren't that difficult to extend pipeing to a general = *clean* concept. I never used pipeing on VMS but I wonder howo> the problem of matching or mapping the outcome of the previous> program to the input of the following program has been solved.6 Maybe another reader can help. Especially SEARCH is an interesting example.  g: > > Why does "ls" offer sorting options - couldn't that beD > > done with pipeing and sort (should be srt!)? Why does "ls" offerA > > output format selection options - couldn't that be done with r@ > > pipeing and grep (should be grp!)? Wouldn't it be much more B > > efficient if the "ls" like the VMS DIRECTORY command could do C > > all sorts of sorting and selecting because it knows best how tot > > interpret the data?r > I > I am the author of the VMS DIRECTORY command.  I have anticipated everya2 > possible output format any user could ever want. > H > Common options are given by command line (or else commands would startE > to look like sausage links).  Pipeing is useful for doing somethingh> > beyond the command line options.  It gives more flexibility.  = This is a half thought through approach. Let's take a look ate: the "dir" command (ls or DIRECTORY). There is a known file@ system. The potential columns of the output are therefore known.: There is limited set of columns which can be displayed and8 according to them could be sorted. I'm not talking about@ different display formats, although DIRECTORY does a fairly good< job by also offering options to define each column width! Of@ course we don't know if the user wants to format the date column@ according to old Chinese animal calender. And DIRECTORY could do< better in offering a parameter to define the ordering of the= columns. But what the ls command offers is simply  redicilous: little and not consistent.  @ Pipeing gives more flexibility, agreed. But it is a productivity@ killer if it is almost every time necessary because the commands: itself are that poor. Pipeing looks like an excuse for the9 system command programmers to not properly do their jobs.   H > You can open a text file with any editor.  There are comments in theseG > text files that explain the configuration.  When you make changes youe0 > can add comments like when, why, by whom, etc.  = Good point taken. Two small objections nevertheless. CommentsB; are very good but not all people put them in. And you couldw? provide comments also in an RMS key file if you design the filee9 that way. I personally recommend using a tool for editingo7 configuration files instead of an editor because of the @ possibility to check the values. A compromise would be to have a? free text configuration file which is then compiled. This givesg> you the best of both worlds. Unfortunately because of the lack> of structured files in UNIX this compromise has many drawbacks@ because you either have to resort to a real DB or you compile to: binary files which are then unreadable by human beeings as@ opposed to RMS files which can still be analysed with the system tools available.@ The problem with the UNIX approach is that the program has to do# the configuration file "compiling".   F > The programmer can create and comment the initial configuration fileI > with a text editor.  There is no need to have some program ask the userdJ > for the configuration and then write it to the keys.  You either have to0 > write the parser or the configuration program.  = You are completely right. But as mentioned before the programe; itself is parsing the configuration file anyway. So why not > moving this code to separate program? The advantage is that as; long as the configuration is not changed each program start ? would be faster and the program would be smaller. Wouldn't thatS be the *real* UNIX approach?  J > By the way, I'm getting the impression that I'm pissing some people off.E > This is not my intention and I'm sorry if I've done so.  I was justi > responding to you statement   @ You are *not* pissing me off. If you are a UNIX afficionado then? fine with me. If you know enough about VMS and computer sciences= you are wellcome because these people are rare. Let me put ite? this way: if you know more about VMS than I about UNIX you will 2 teach me something I want to learn and understand.   > $ > > Linux has the old, obsolete UNIX? > > architecture and is by that like all other UNIXes much lessd > > modern than VMS. > @ > > Of course I'm on my way back to VMS and I will do as much asA > > possible on VMS because the architecture is clean in every ofhB > > the three important dimensions. And the UNIX architecture will6 > > never be clean; if it will it is not UNIX anymore. > 5 > because I feel that Unix is the cleaner of the two.    Feeling is not knowing ...  J > I like VMS, I just like Linux more.  It was not my intention to slam VMS( > and I'm sorry if it appeared that way. >  > No hard feelings.c >  > Fran >  > Sent via Deja.coma > http://www.deja.com/  @ It didn't appear that way to me. I still want to understand what> the *real* advantages of UNIX are and why Linux is such a hype= although it seems that there is kind of understanding in this 0 thread that it is even not the best UNIX around.   What do you like with VMS?   ------------------------------  # Date: Fri, 26 Jan 2001 22:13:34 GMTs2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)& Subject: Re: VMS writable CD on Win2K?7 Message-ID: <iEmc6.142$cu.1115@gazette.loc1.tandem.com>u  p In article <009F6A96.208B4B9F@SendSpamHere.ORG>, system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) writes:Z :In article <94qb5p$8nsg$1@lead.zk3.dec.com>, "Paul A. Jacobi" <nospan@nospam.com> writes: :>L :>"Brian Schenkenberger, VAXman-" <system@SendSpamHere.ORG> wrote in message, :>news:009F69C2.220C20D2@SendSpamHere.ORG... :>>m+ :>> Curious, what do you use for write s/w?a :>K :>Uh, unfortunately, we had to use a PC with a DVD-R drive to write the ODSh
 :>image file.i :>L :>We are aware of several request to officially support writeable CD and DVDN :>drives under OpenVMS systems, but it has not been determined if/when we will :>ever provide the support.  ..K :Excuse me but didn't you post telling another person that you have various % :CD-R attached to VMS and in use???  w  H   And yes, we do.  (We've been working with directly using DVD-R in the G   local hunk of OpenVMS Engineering, DVD media is just large enough to g0   archive the results of OpenVMS system builds.)  J   CD-R and DVD have been working for a while now, with the initial work on#   DVD in an IDE ECO for V7.1-1H2.  d  H   That said, Compaq OpenVMS Management has not yet formally committed toI   provide SPD-level support, and has not chosen any particular CD-R, DVD  1   or DVD-R drive(s) to formally test and support.   J   Folks at US Design do have a formally supported configuration available E   for CD-R operations on OpenVMS.  A pointer is available in the FAQ.e  J :Also, there is no reason to have to use a PC to write the ODS disk image.; :See http://www.tmesis.com/CDrom/ for some pointers to s/w.h  D   We have been writing directly to the device, and using versions or*   variants of tools referenced in the FAQ.  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Sat, 27 Jan 2001 00:28:51 GMTa= From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-)e& Subject: Re: VMS writable CD on Win2K?0 Message-ID: <009F6B56.AD30CFFC@SendSpamHere.ORG>  l In article <iEmc6.142$cu.1115@gazette.loc1.tandem.com>, hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) writes:q >In article <009F6A96.208B4B9F@SendSpamHere.ORG>, system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) writes:l[ >:In article <94qb5p$8nsg$1@lead.zk3.dec.com>, "Paul A. Jacobi" <nospan@nospam.com> writes:i >:>cM >:>"Brian Schenkenberger, VAXman-" <system@SendSpamHere.ORG> wrote in messagea- >:>news:009F69C2.220C20D2@SendSpamHere.ORG...i >:>>, >:>> Curious, what do you use for write s/w? >:>pL >:>Uh, unfortunately, we had to use a PC with a DVD-R drive to write the ODS >:>image file. >:>*M >:>We are aware of several request to officially support writeable CD and DVDkO >:>drives under OpenVMS systems, but it has not been determined if/when we willt >:>ever provide the support. >...L >:Excuse me but didn't you post telling another person that you have various& >:CD-R attached to VMS and in use???   >cI >  And yes, we do.  (We've been working with directly using DVD-R in the yH >  local hunk of OpenVMS Engineering, DVD media is just large enough to 1 >  archive the results of OpenVMS system builds.)p >eK >  CD-R and DVD have been working for a while now, with the initial work onw$ >  DVD in an IDE ECO for V7.1-1H2.   >PI >  That said, Compaq OpenVMS Management has not yet formally committed tooJ >  provide SPD-level support, and has not chosen any particular CD-R, DVD 2 >  or DVD-R drive(s) to formally test and support. > K >  Folks at US Design do have a formally supported configuration available eF >  for CD-R operations on OpenVMS.  A pointer is available in the FAQ.   Which is CDrecord.   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMr            nO city, n., 1. a place where trees are cut down and streets are named after them.    ------------------------------  # Date: Fri, 26 Jan 2001 21:42:56 GMTo" From: fooguy <jweisen@my-deja.com>9 Subject: Volume Shadowing vs. RA7000 (WAS: DS20 Slow I/O)t) Message-ID: <94sr0v$g7v$1@nnrp1.deja.com>t  F So we're sending all this hardware back, but now we *really* need your help.o  A Our proposed alternative config is each system gets (2) KZPCA LVDeH controllers connecting to a 4254 U160 Disk Shelf with about (14) 18G 10K= RPM Disks. The catch is that the LVD controller is not a RAID F controller, so we'd have to use Volume Shadowing. We propose mirroringG the disks across controllers (i.e DSA0: is composed of DKA0: and DKB0:)GG for a full duplex software mirror.  In this scenerio, the controller isg. 80MB/Sec, the Cabinet and Disks are 160MB/Sec.  G Our vendor doesn't like the idea of soft mirror, and wants us to get an D RA7000 with (2) HSZ70s. In this config, both systems would share theA RA7000, so each would be connected to one of the HSZ70s. Now, theaD controller in the system, the HSZ70, and the storage works disks are- Ultra, so they have a max rating of 40MB/Sec.i  G What we're wondering is, how much overhead does Volume Shadowing reallyaE generate? I'm on hold as we speak with Compaq to do a 30 day trial of E the Vol Shadow license so we can try it out, but your experience withmD the HSZ70, the RA7000, and/or Volume Shadowing would be appreciated.   Thanks,  John   --- *********************************************t( "All I every wanted from life was to see, Larry Wall give Bill Gates a Perl Necklace."     Sent via Deja.coma http://www.deja.com/   ------------------------------    Date: 26 Jan 2001 20:10:29 -0500* From: young_r@eisner.decus.org (Rob Young)= Subject: Re: Volume Shadowing vs. RA7000 (WAS: DS20 Slow I/O)t+ Message-ID: <tRRcIHYAkxMq@eisner.decus.org>c  N In article <94sr0v$g7v$1@nnrp1.deja.com>, fooguy <jweisen@my-deja.com> writes:H > So we're sending all this hardware back, but now we *really* need your > help.t > C > Our proposed alternative config is each system gets (2) KZPCA LVDaJ > controllers connecting to a 4254 U160 Disk Shelf with about (14) 18G 10K? > RPM Disks. The catch is that the LVD controller is not a RAIDhH > controller, so we'd have to use Volume Shadowing. We propose mirroringI > the disks across controllers (i.e DSA0: is composed of DKA0: and DKB0:)vI > for a full duplex software mirror.  In this scenerio, the controller ise0 > 80MB/Sec, the Cabinet and Disks are 160MB/Sec. >   / 	You have a great winner with volume shadowing.  	 : 	You can get volume shadowing and as I pointed out earlier9 	pick up RAID Software for OpenVMS ($1000 list price) andi. 	do striping across two or three shadowsets.    . 	David Mathog is right.  You can use a RAMDISK8 	for I/O.  The RAMDISK can be one member of a shadowset!6 	You don't lose writes that way in case of power loss!    I > Our vendor doesn't like the idea of soft mirror, and wants us to get an F > RA7000 with (2) HSZ70s. In this config, both systems would share theC > RA7000, so each would be connected to one of the HSZ70s. Now, theoF > controller in the system, the HSZ70, and the storage works disks are/ > Ultra, so they have a max rating of 40MB/Sec.g >   / 	He doesn't like the "idea" of a "soft mirror?"f9 	Betcha you he can add his VMS knowledge up and wouldn't  5 	have enough to blow his nose.  Ask him for technicalt& 	details in a none threatening manner.  I > What we're wondering is, how much overhead does Volume Shadowing really G > generate? I'm on hold as we speak with Compaq to do a 30 day trial oflG > the Vol Shadow license so we can try it out, but your experience witheF > the HSZ70, the RA7000, and/or Volume Shadowing would be appreciated. >   = 	Volume Shadowing uses so little overhad, I can't measure it.n; 	Remember that the kind of thing that generates overhead isr= 	RAID5.  RAID5 has to write parity information and the paritye9 	information has to be calculated, etc.  Volume Shadowingf@ 	decides (via queue length info handed back from the controller); 	which unit is least busy and targets the read to the least @ 	busy unit.  That kind of overhead is very lightweight.  PerhapsB 	someone *really* in the know can comment as to percentages... but: 	I'll tell you this.. I can't see it affecting my systems.   	More later...   				Rob    ------------------------------   Date: 26 Jan 2001 17:56:43 PSTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)= Subject: Re: Volume Shadowing vs. RA7000 (WAS: DS20 Slow I/O) 3 Message-ID: <Xku5qArpVwCF@mccdev.slac.stanford.edu>o  , In article <tRRcIHYAkxMq@eisner.decus.org>, 1     	young_r@eisner.decus.org (Rob Young) writes:m [...] ? > 	Volume Shadowing uses so little overhad, I can't measure it.r  D         To add a smidgen to what Rob said, reads can be faster underD     HBVS because you have two (at least) members to read from so you;     can get the data off the least busy platter/controller.n  H         On the  other  side,  writes  must  be  executed  on  all shadowH     members.   That is, two (or three) writes are queued by VMS, one  toH     each shadow member.  But VMS  does  asynchonous  I/O  so  queuing  a+     second write adds very little overhead.y           -Ken --  M  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edux:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  % Date: Fri, 26 Jan 2001 09:56:24 +0000s% From: Alan Greig <agreig@my-deja.com>b6 Subject: Re: Window for Alpha success still seems open8 Message-ID: <67i27tchn0m5ia7jek38g0pvngmbk1qgdo@4ax.com>  = On 25 Jan 2001 13:40:21 -0500, jordan@lisa.gemair.com (Jordanh Henderson) wrote:a   >Take a look at this article:? >o& > http://www.tweakers.net/reviews/204/ >r? >It seems that the Itanium is benchmarking unbelievably poorly,I< >especially when executing 32 bit code.  A 100Mhz Pentium I - >outbenches a 1.5 Ghz Itanium in some cases!?e  C Actually it was a 667 Mhz Itanium that performed like a P100. StillaE disastrous thougjh I would say. Also they couldn't actually benchmarkh? any 64 bit code but commented that performance of 64bit windowsxA indicated that it was at least as fast as a Pentium 4 at the samei
 clock speed .c     --
 Alan Greig   ------------------------------    Date: 26 Jan 2001 05:58:00 -0500/ From: jordan@lisa.gemair.com (Jordan Henderson)t6 Subject: Re: Window for Alpha success still seems open* Message-ID: <94rl7o$hl8$1@lisa.gemair.com>  8 In article <67i27tchn0m5ia7jek38g0pvngmbk1qgdo@4ax.com>,' Alan Greig  <agreig@my-deja.com> wrote:h> >On 25 Jan 2001 13:40:21 -0500, jordan@lisa.gemair.com (Jordan >Henderson) wrote: >r >>Take a look at this article: >>' >> http://www.tweakers.net/reviews/204/i >>@ >>It seems that the Itanium is benchmarking unbelievably poorly,= >>especially when executing 32 bit code.  A 100Mhz Pentium I h. >>outbenches a 1.5 Ghz Itanium in some cases!? >rD >Actually it was a 667 Mhz Itanium that performed like a P100. StillF >disastrous thougjh I would say. Also they couldn't actually benchmark@ >any 64 bit code but commented that performance of 64bit windowsB >indicated that it was at least as fast as a Pentium 4 at the same >clock speed . >   9 Oh, yeah, you're right.  They only had a 667 Mhz Itanium.   D The thing is that perceptions are going to be formed by how well theB IA-64s run IA-32 code, because that's all that customers currently+ have initially, their current applications.m   >a >--  >Alan GreigR   -Jordan Hendersonv jordan@greenapple.com    ------------------------------  % Date: Fri, 26 Jan 2001 11:57:07 +0000g% From: Alan Greig <agreig@my-deja.com>w6 Subject: Re: Window for Alpha success still seems open8 Message-ID: <3sn27t8euhg6foguu515q0t54os1n2vrg5@4ax.com>  = On 26 Jan 2001 05:58:00 -0500, jordan@lisa.gemair.com (Jordanh Henderson) wrote:k     >k: >Oh, yeah, you're right.  They only had a 667 Mhz Itanium. >eE >The thing is that perceptions are going to be formed by how well therC >IA-64s run IA-32 code, because that's all that customers currentlyt, >have initially, their current applications.  F Here's a real life example of how I will judge it. We use Alpha/NT forF one thing only: to run ANSYS. In certain large scale models an EV6 500* Mhz Alpha performs like a 2 Ghz Pentium 4.  E We won't be buying IA64 desktops in the forseeable future but we willrE benchmark an IA64 with Ansys as soon as one is available. Now I guessa@ it's the case that most vendors who had Alpha/NT ports will haveD IA64/NT ports available close to public launch - ANSYS already listsD the IA64 version as *shipping* for both IA64/HP-UX and IA64/NT 2000.? It is the performance of these apps that IA64 lives or dies by.y@ Unfortunately the Ansys site gives no benchmark figures as yet -@ presumably for legal reasons. Alpha tops the currently published figures.  ? I personally don't give a damn about how fast it as at IA32 and < neither do I suspect do most of its initial target audience.   --
 Alan Greig   ------------------------------  % Date: Fri, 26 Jan 2001 15:22:26 +0000 - From: Tim Llewellyn <tim.llewellyn@bbc.co.uk> 6 Subject: Re: Window for Alpha success still seems open( Message-ID: <3A719632.5F33445@bbc.co.uk>   Alan Greig wrote:n  ? > On 26 Jan 2001 05:58:00 -0500, jordan@lisa.gemair.com (Jordan. > Henderson) wrote:e >  > >a< > >Oh, yeah, you're right.  They only had a 667 Mhz Itanium. > > G > >The thing is that perceptions are going to be formed by how well theuE > >IA-64s run IA-32 code, because that's all that customers currentlyr. > >have initially, their current applications. >aH > Here's a real life example of how I will judge it. We use Alpha/NT forH > one thing only: to run ANSYS. In certain large scale models an EV6 500, > Mhz Alpha performs like a 2 Ghz Pentium 4. >s  I Presumably that app isn't parallelizable, or it would be a lot cheaper ton get 4 PIII 1GHz's.    --g6 Tim Llewellyn, OpenVMS Infrastructure, Remarcs Project0 MedAS at the BBC, Whiteladies Road, Bristol, UK.A Email tim.llewellyn@bbc.co.uk. Home tim.llewellyn@cableinet.co.uk   A I speak for myself only and my views in no way represent those of  MedAS or the BBC.f   ------------------------------  % Date: Fri, 26 Jan 2001 16:21:49 +0000o% From: Alan Greig <agreig@my-deja.com>m6 Subject: Re: Window for Alpha success still seems open8 Message-ID: <o6837t47cbmetpvjrc3rjjojjrg16taumb@4ax.com>  1 On Fri, 26 Jan 2001 15:22:26 +0000, Tim Llewellynl  <tim.llewellyn@bbc.co.uk> wrote:   >e >n >Alan Greig wrote: >'@ >> On 26 Jan 2001 05:58:00 -0500, jordan@lisa.gemair.com (Jordan >> Henderson) wrote: >> >> >= >> >Oh, yeah, you're right.  They only had a 667 Mhz Itanium.f >> >H >> >The thing is that perceptions are going to be formed by how well theF >> >IA-64s run IA-32 code, because that's all that customers currently/ >> >have initially, their current applications.V >>I >> Here's a real life example of how I will judge it. We use Alpha/NT fornI >> one thing only: to run ANSYS. In certain large scale models an EV6 500t- >> Mhz Alpha performs like a 2 Ghz Pentium 4.n >> >MJ >Presumably that app isn't parallelizable, or it would be a lot cheaper to >get 4 PIII 1GHz's.a  E Ansys will take advantage of multiple processors if it can and I havesE had this discussion with our design engineers and they with Ansys. It ? appears that for our really long running calculations we cannoteF parallelize to any significant degree. What I've been told is that theC stages that run in 4 minutes could run in 1 on a 4 processor systemf= but the bit that takes 4 days would also run in 4 days on a 4 A processor system.  We're talking large under-sea assemblies here.t  A A couple of months ago one of our senior engineers took a trip to @ Moscow University's dept of Mathematics to see if we could speed; things up; apparently they have some of the best experts innD computational finite element analysis Their answer was no. Not if we wanted to be sure anyway.e   --
 Alan Greig   ------------------------------   Date: 26 Jan 2001 17:28:05 GMT2 From: mathog@seqaxp.bio.caltech.edu (David Mathog)6 Subject: Re: Window for Alpha success still seems open, Message-ID: <94sc35$b7o@gap.cco.caltech.edu>  ` In article <67i27tchn0m5ia7jek38g0pvngmbk1qgdo@4ax.com>, Alan Greig <agreig@my-deja.com> writes:> >On 25 Jan 2001 13:40:21 -0500, jordan@lisa.gemair.com (Jordan >Henderson) wrote: >- >>Take a look at this article: >>' >> http://www.tweakers.net/reviews/204/a >>@ >>It seems that the Itanium is benchmarking unbelievably poorly,= >>especially when executing 32 bit code.  A 100Mhz Pentium I r. >>outbenches a 1.5 Ghz Itanium in some cases!? > ? >Actually it was a 667 Mhz Itanium that performed like a P100. v  E For a while I ran a 200 4/166 alphastation with NT and put some intel K programs in FX!86.  Those ran at about the speed of Pentium 60.  Scaling uphK to a 600 Mhz Alpha (straight clock speed), and allowing another factor of 3 J for the 21264 over the older chip, it should now run like a 3*60*600/166 =I 650 Mhz pentium in FX!86.   Except that you can't get Windows of any typehH for Alpha anymore, nor FX!86.  But the emulation is only important untilH the native apps are released.  Digital was incredibly STUPID not to haveK locked MS into a contract to provide native MS Office for WNT/Alpha.  IntelmI will not make that mistake. And they may not need to do much to get that -J crucial piece of monopolyware, since I think it likely that Microsoft willJ see the Itanium as the future home for 64 bit Windows and do the port with little or no prodding. p  G And there we see the difference between those who know how to win, and e those who do not.O   Sigh.m   David Mathog mathog@seqaxp.bio.caltech.edue? Manager, sequence analysis facility, biology division, Caltech c   ------------------------------    Date: 26 Jan 2001 13:14:05 -0500/ From: jordan@lisa.gemair.com (Jordan Henderson)i6 Subject: Re: Window for Alpha success still seems open* Message-ID: <94sepd$erh$1@lisa.gemair.com>  8 In article <3sn27t8euhg6foguu515q0t54os1n2vrg5@4ax.com>,' Alan Greig  <agreig@my-deja.com> wrote:S> >On 26 Jan 2001 05:58:00 -0500, jordan@lisa.gemair.com (Jordan >Henderson) wrote: >e >g >>; >>Oh, yeah, you're right.  They only had a 667 Mhz Itanium.4 >>F >>The thing is that perceptions are going to be formed by how well theD >>IA-64s run IA-32 code, because that's all that customers currently- >>have initially, their current applications.- >rG >Here's a real life example of how I will judge it. We use Alpha/NT fornG >one thing only: to run ANSYS. In certain large scale models an EV6 500t+ >Mhz Alpha performs like a 2 Ghz Pentium 4.w >0F >We won't be buying IA64 desktops in the forseeable future but we willF >benchmark an IA64 with Ansys as soon as one is available. Now I guessA >it's the case that most vendors who had Alpha/NT ports will have E >IA64/NT ports available close to public launch - ANSYS already listseE >the IA64 version as *shipping* for both IA64/HP-UX and IA64/NT 2000.t@ >It is the performance of these apps that IA64 lives or dies by.A >Unfortunately the Ansys site gives no benchmark figures as yet -oA >presumably for legal reasons. Alpha tops the currently publishedi	 >figures.  >n@ >I personally don't give a damn about how fast it as at IA32 and= >neither do I suspect do most of its initial target audience.t >b  ? The initial target audience may be irrelevant.  People will gett; IA64 for their systems in order to have the thing with the  : highest numbers in it's name, much like people buy 1.4 Ghz< Pentium 4's now that don't perform any better than a 1.2 Ghz) Althlon for their particular application.-  ; Once the word is out that the IA64 is only for specialized s; markets, like your own, then there won't be much commercial ; demand for them.  Lack of demand could translate to lack ofh= development money.  Remember that Intel can't milk the Wintelc( market like they used to because of AMD.  = If Alpha maintains a significant lead on IA64, and IA32, thent there may be some hope.e  > I do agree with David Mathog's evaluation of what has happened to Alpha in the past...r   >--g >Alan Greigy   -Jordan HendersonV jordan@greenapple.com0   ------------------------------  # Date: Fri, 26 Jan 2001 18:47:21 GMTo. From: Cayemberg, Keith <cayemberg@my-deja.com>6 Subject: Re: Window for Alpha success still seems open) Message-ID: <94sgnm$66l$1@nnrp1.deja.com>   ( In article <3A719632.5F33445@bbc.co.uk>,    tim.llewellyn@bbc.co.uk wrote: >k >T > Alan Greig wrote:L >c   ...e ...n  F > > Here's a real life example of how I will judge it. We use Alpha/NT foriF > > one thing only: to run ANSYS. In certain large scale models an EV6 500 . > > Mhz Alpha performs like a 2 Ghz Pentium 4. > >  > H > Presumably that app isn't parallelizable, or it would be a lot cheaper to > get 4 PIII 1GHz's.  < I would say it's not just a matter of whether the problem is9 parallelizable. Important is also the degree to which thec> parallelized solution has dependencies requiring communication2 and/or synchronization between the parallel tasks.: If such synchonization/communication is necessary then the; 4 PIII's communicating over a network will never come closegA to a 4 CPU system with a switched bus design. Nor close to even ad? 1 cpu design that has a significant percentage of the computingu7 power of the 4 metworked PC's (which the Alpha offers).S  = However, if you make a 4 PIII SMP System with a switched bus,*8 the resulting system will cost significantly more than 4; separate networked commodity PC's. It might cost as much asa0 an ES40, without the ES40's stellar performance!  9 Not every problem fits the Beowulf Cluster or SETI model.:    : As for the Itanium and it's long announced future dynasty.< The only reason why it might have some success, is if people: believe in the marketing. It's clear to me that Intel made9 the wrong decisions with it's EPIC VLIW design. The Alpha ; design team considered making an Alpha with the same design*9 principles back in 1993. Through initial prototyping they : decided that it's a dead-end technology, which effectively8 blocks the use of known and future dynamic optimizations< for the sake of static optimizations which can never make-up8 for the ability adjust dynamically to problem situation.  > Note, that this viewpoint was clear to designers which alreadyE had a relatively simple chip architecture (still without complexitiesc8 such as out-of-order processing) inwhich to base such an excursion into VLIW design.u  > You can read about the Alpha design teams take on RISC vs VLIW here...w  4 http://www.alphapowered.com/alpha_tech_presents.html  - Look for the document titled Alpha and IA-64.g    < For an interesting comparison of the pro's and con's of SMT,= CMP, and VLIW prlease read the recent articles by Paul DeMone * about Alpha at Silicon Insider RWT here...   http://www.realworldtech.com    ? I find a huge waste Intel's, HP's, and their business partner'sa6 money, resources and talent to invest in such a poorly5 considered design. Worse, is that competitors gave upd9 their perfectly good architectures, simply due to Intel'sm@ marketing strength without considering the real (lack of) merits of it's technology.B  9 As long as Compaq continues on the course it has outlinedf; for Alpha, and doesn't get stuck implementing the SMT core, ; there is no chance IA-64 will ever get close to beating thei, Alpha Architecture. The Intel designers have> simply handicapped themselves too much with their basic design9 decisions. For instance, SMT will be nearly impossible to09 implement based on a IA-64 core. And, it will necessarilyu: run hotter with more timing hurdles than Alpha at the same> process shrink, thus limiting it's frequency range. That's one> of the reasons Itanium started with a 1Ghz 1st generation spec: 3 years ago, reduced later to 800 Mhz, and now is settling: for only 733 Mhz. The VLIW design explodes the real estate! needed to implement a 64bit chip.m  2 If Intel wants to beat Alpha, they've got to start? all over again and design a different architecture. Maybe theirh6 marketing bozo's would call it IA-64 too, but it would= remain clear to us techies that it's a different chip family,a! and that Intel failed completely.    Regards,   Keith Cayemberge     Sent via Deja.comm http://www.deja.com/   ------------------------------  # Date: Sat, 27 Jan 2001 02:20:20 GMT3/ From: StevenU@POBoxes.com (Steven P. Underwood)f6 Subject: Re: Window for Alpha success still seems open2 Message-ID: <3a722f24.177768889@news.telocity.com>  A On 26 Jan 2001 17:28:05 GMT, mathog@seqaxp.bio.caltech.edu (Davidc Mathog) wrote:  F >For a while I ran a 200 4/166 alphastation with NT and put some intelL >programs in FX!86.  Those ran at about the speed of Pentium 60.  Scaling upL >to a 600 Mhz Alpha (straight clock speed), and allowing another factor of 3K >for the 21264 over the older chip, it should now run like a 3*60*600/166 =tJ >650 Mhz pentium in FX!86.   Except that you can't get Windows of any typeI >for Alpha anymore, nor FX!86.  But the emulation is only important untilw  B While I agree with your basic ideas, I think your wording is a bitD "loose".  You can still get WinNT v4.0 which still runs on the olderB Alphas and you can still download and run FX!32 from the followingC Compaq web site.  http://www.support.compaq.com/amt/fx32/index.html   B I believe you were trying to say that Windows will not work on theE newest Alpha's, which I think is true, but would never attempt such a-
 downgrade.   Steve-   >- >Sigh. >e
 >David Mathog  >mathog@seqaxp.bio.caltech.edu@ >Manager, sequence analysis facility, biology division, Caltech    Steven P. Underwood,DNRC Whitinsville,MAg StevenU@POBoxes.comu   ------------------------------  % Date: Thu, 25 Jan 2001 22:54:27 -0800n! From: Koloth <koloth@tmisnet.com> 0 Subject: Re: Writes flushed to Disk on Dismount?+ Message-ID: <3A711F22.8D955755@tmisnet.com>    Can someone tell me why raising the cache flush timer to 45 or 64K seconds improve performance?  Is there some guidance for selecting ae( value?  Is there a way to  measure this?   Cass   ------------------------------    Date: 26 Jan 2001 09:42:48 +0100G From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de>o0 Subject: Re: Writes flushed to Disk on Dismount?H Message-ID: <y4y9vyn1pz.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>  3 "Dave Gudewicz" <david.gudewicz@abbott.com> writes:t  L > When you dismount the disk on an HSJ controller, cache is flushed to disk.L > On an HSZ or HSG controller, the data remains in cache. The reason is thatF > the HSJ uses the MSCP protocol, and a dismount translates to an MSCPL > AVAILABLE command. This command results in a cache flush. On a G or Z, the1 > protocol is SCSI, and there is no such command.>  E I can understand that SCSI doesn't have a straight equivalent to MSCP E AVAILABLE. But it seems reasonable that there should be some command wF that tells the drive to flush its cache, or possibly one that disablesD it and as a side effect flushes any still remaining cache contents. H The HSZ/G port driver (whatever it is called today) should then be able G to translate an AVAILABLE to a series of SCSI commands containing that.s   	Jan   ------------------------------    Date: 26 Jan 2001 10:11:35 -0500* From: young_r@eisner.decus.org (Rob Young)0 Subject: Re: Writes flushed to Disk on Dismount?+ Message-ID: <bBT2+bDH9kQr@eisner.decus.org>   l In article <94q6uu$bs5$1@fizban.fizban.pprd.abbott.com>, "Dave Gudewicz" <david.gudewicz@abbott.com> writes: > 9 > "Rob Young" <young_r@eisner.decus.org> wrote in message ' > news:a$agcRuKEHCF@eisner.decus.org...lK >> In article <94puu9$bpm$1@fizban.fizban.pprd.abbott.com>, "Dave Gudewicz"o% > <david.gudewicz@abbott.com> writes:- >> >> > >> >3 >> > I'll ask KB and let you all know.  How's that.  >> > >> > Dave... >> > >>G >> HA!  Cat's out of the bag.  Wonderful being in this tight community.3 >>> >> The boys and girls that travel with carnivals must have the? >> same warm fuzzy feeling at times.  Wonder if they pack along A >> a mirror copy of our British friend.  Nahhhh.. they would have A >> all hopped in the trucks at the end of the show and peeled outh >> without saying a word.  >>  >> Thanks for looking into that. >> >> Rob >> > 5 > KB says:  Works different on the HSJ Vs. the HSZ/G.e >  > The HSZ/G story: > L > If you do this, you should issue a "set d1 norun" HSx command to make sure
 > that things0I > are cleaned up. A byproduct of this command is that it flushes cache tok > disk.h >  > The HSJ story: > L > When you dismount the disk on an HSJ controller, cache is flushed to disk.L > On an HSZ or HSG controller, the data remains in cache. The reason is thatF > the HSJ uses the MSCP protocol, and a dismount translates to an MSCPL > AVAILABLE command. This command results in a cache flush. On a G or Z, the1 > protocol is SCSI, and there is no such command.t >  > Hoped I helped out a little. >   6 	This helps out a lot.  In the future, I won't ( until5 	scripts are modified and tested ) be jacking up the n7 	CACHE_FLUSH_TIMER or all my backups will be backing up   	inconsistent shadowset members.  = 	To work around this, it requires the _____ (HSDSA equivalent-5 	can't recall name but is on my list) utility to firei9 	controller commands to "set norun" and then "set run" on < 	reintegration.  Sounds like the kind of kludge that doesn't 	make me giddy.    				Robw   ------------------------------    Date: 26 Jan 2001 10:27:12 -0500* From: young_r@eisner.decus.org (Rob Young)0 Subject: Re: Writes flushed to Disk on Dismount?+ Message-ID: <1cNW72S$kw6Z@eisner.decus.org>a  l In article <94q6uu$bs5$1@fizban.fizban.pprd.abbott.com>, "Dave Gudewicz" <david.gudewicz@abbott.com> writes:   > 5 > KB says:  Works different on the HSJ Vs. the HSZ/G.a >  > The HSZ/G story: > L > If you do this, you should issue a "set d1 norun" HSx command to make sure
 > that thingsdI > are cleaned up. A byproduct of this command is that it flushes cache tob > disk.p >  > The HSJ story: > L > When you dismount the disk on an HSJ controller, cache is flushed to disk.L > On an HSZ or HSG controller, the data remains in cache. The reason is thatF > the HSJ uses the MSCP protocol, and a dismount translates to an MSCPL > AVAILABLE command. This command results in a cache flush. On a G or Z, the1 > protocol is SCSI, and there is no such command.  >  > Hoped I helped out a little. >   . 	Here is the follow-up that just dawned on me.  < 	If you dismount that drive to break it out of the shadowset< 	and remount it to backup... will it hit the same controller  	in a dual-redundant HSZ pair?    < 	If running mirrored cache in HSG , my concern may not be an 	issue.   7 	If you dismount/mount and then perform backup, yes you = 	may have writes that hadn't been committed.  But when Backupe= 	comes along and goes to read that LBN that is in cache, thatc< 	is just what it does.  So Backup probably gets a consistent 	view.  6 	In some ways, I come to realize I (once again) may be? 	making a Mountain out of a mole-hill  (however, for folks withf? 	HSZ and/or running HSG with split cache the real concern wouldi; 	be if on remount and no preferreed path it hits the OTHER 0: 	controller that would indeed NOT of the blocks that have ; 	been modified.. so maybe a large hill out of a mole-hill).o   				Robo   ------------------------------  # Date: Fri, 26 Jan 2001 15:36:15 GMTs$ From: Scott Vieth <svieth@wi.rr.com>0 Subject: Re: Writes flushed to Disk on Dismount?& Message-ID: <3A7199C2.60609@wi.rr.com>   Hey Ken:  B Haven't chatted with you since we were working on stuffing the 4k  firmware into CIXCDs....  7 Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515 wrote:s  J >         But isn't the real issue  whether  you  get all writes completedJ >     and  flushed  to  disk on system shutdown (disk  dismount)  so  thatJ >     you're assured you can power things down and have the  data  intact?  H Are you talking about powering down your hosts or the storage cabinet?   If it's the storage cabinet,I aren't you going to do a "SHUTDOWN OTHER" and "SHUTDOWN THIS" before you t powereI down your storage array?  I bet *those* commands would flush your writes   to disk.  I Why would you pull a drive out of your storage cabinet (even if they are R hot-swappable) without deleting  the unit and the disk first?  J >     I  dunno,  having  to  worry whether the SCSI disk I just dismountedJ >     really had all its  data  written  before  swapping a controller, or+ >     something, makes me pretty nervous...s  F Now you're talking about run CSWAP (or whatever the name of the board  swap utility is.  I haven'tsG used it in a a while....)  Again, I bet that executing the commands to w) replace a controller board in a redundantm pair would flush the caches.  M >> I've been out-of-the-loop for a while.  Who's KB?  Is it a secret?  If so,eG >> email me offline.  You can trust me to keep your secret.  Ha ha ha!!s >  > @ >         Wouldn't that be Ken Burns, storage guru extrodinaire?  @ Never heard of him.  Is he related to C. Montgomery Burns?   ;^)   -Scott in Brewcity  :^)o   ------------------------------  # Date: Fri, 26 Jan 2001 15:39:47 GMT $ From: Scott Vieth <svieth@wi.rr.com>0 Subject: Re: Writes flushed to Disk on Dismount?( Message-ID: <3A719A97.6090803@wi.rr.com>  G My best guess is that if you had an application doing a ton of writes, T the controllers could cachenF a lot of blocks without having to start writing them to disk after 45  seconds.  By setting theF timeout to 64k second, you are telling the controller "Write these to  the disk when you get a chance.i# There's no hurry.  Take your time."f  C The only good way to measure the performance increase is to slowly n increase the number (you canI monkey with it on the fly) and see if you apps run quicker.  But I think i that the max_cached_transfer, size would also be worth looking at as well.  
 -Scott :^)  
 Koloth wrote:,   > Can someone tell me why raising the cache flush timer to 45 or 64K seconds improve performance?  Is there some guidance for selecting aR* > value?  Is there a way to  measure this? >  > Cass >  >  >    ------------------------------    Date: 26 Jan 2001 10:40:12 -0500* From: young_r@eisner.decus.org (Rob Young)0 Subject: Re: Writes flushed to Disk on Dismount?+ Message-ID: <lxRhPQqJWyCN@eisner.decus.org>e  M In article <3A7199C2.60609@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes: 
 > Hey Ken: > D > Haven't chatted with you since we were working on stuffing the 4k  > firmware into CIXCDs.... > 9 > Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515 wrote:o > K >>         But isn't the real issue  whether  you  get all writes completednK >>     and  flushed  to  disk on system shutdown (disk  dismount)  so  thatMK >>     you're assured you can power things down and have the  data  intact?t > J > Are you talking about powering down your hosts or the storage cabinet?   > If it's the storage cabinet,K > aren't you going to do a "SHUTDOWN OTHER" and "SHUTDOWN THIS" before you T > poweraK > down your storage array?  I bet *those* commands would flush your writes h
 > to disk. >   9 	In earlier discussions I have been told shutdown flushess= 	writes.  That would on the surface appear pretty obvious butt= 	if it isn't documented would make one nervous thinking aboute 	things like that.   				Rob    ------------------------------  # Date: Fri, 26 Jan 2001 15:47:09 GMTt$ From: Scott Vieth <svieth@wi.rr.com>0 Subject: Re: Writes flushed to Disk on Dismount?( Message-ID: <3A719C51.9000107@wi.rr.com>   Rob Young wrote:  8 > 	This helps out a lot.  In the future, I won't ( until7 > 	scripts are modified and tested ) be jacking up the y9 > 	CACHE_FLUSH_TIMER or all my backups will be backing upn" > 	inconsistent shadowset members.  I Why would you say they are inconsistent?  Aren't the blocks still in the z controller's cache?o- (I'm assuming we're talking about HSGs here).e  ? > 	To work around this, it requires the _____ (HSDSA equivalentm7 > 	can't recall name but is on my list) utility to firea; > 	controller commands to "set norun" and then "set run" ona> > 	reintegration.  Sounds like the kind of kludge that doesn't > 	make me giddy.w > 	 > 				Robg  H I haven't yet had the pleasure to connect an HSG80 to a VMS system.  Is % there a program for the VMS/HSG worldeH equivalent to the HSDSA that I used to send commands to HSJ controllers?   thanks much,  
 -Scott :^)   ------------------------------    Date: 26 Jan 2001 11:03:36 -0500* From: young_r@eisner.decus.org (Rob Young)0 Subject: Re: Writes flushed to Disk on Dismount?+ Message-ID: <Xdn5VOHz2cfi@eisner.decus.org>   O In article <3A719C51.9000107@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes:  >  >  > Rob Young wrote: > 9 >> 	This helps out a lot.  In the future, I won't ( untilr8 >> 	scripts are modified and tested ) be jacking up the : >> 	CACHE_FLUSH_TIMER or all my backups will be backing up# >> 	inconsistent shadowset members.e > K > Why would you say they are inconsistent?  Aren't the blocks still in the h > controller's cache?z/ > (I'm assuming we're talking about HSGs here).  > @ >> 	To work around this, it requires the _____ (HSDSA equivalent8 >> 	can't recall name but is on my list) utility to fire< >> 	controller commands to "set norun" and then "set run" on? >> 	reintegration.  Sounds like the kind of kludge that doesn't  >> 	make me giddy. >> n
 >> 				Rob > J > I haven't yet had the pleasure to connect an HSG80 to a VMS system.  Is ' > there a program for the VMS/HSG worldpJ > equivalent to the HSDSA that I used to send commands to HSJ controllers? >  > thanks much, >   8 http://www5.compaq.com/products/sanworks/sanmgt/scs.html      3                          SANworksT Command Scriptere    N The Compaq SANworks Command Scripter application software provides experiencedO IT managers with command-level control of Compaq StorageWorksT systems equippedrO with HSG60, HSG80, HSZ70 and HSZ80 Array Controllers. Command Scripter works insO a heterogeneous host environment that includes Compaq Tru64 UNIX, Windows 2000, % Windows NT, Sun Solaris, and OpenVMS.n  N With Command Scripter, IT managers can create, edit, and run script files thatH contain StorageWorks Command Line Interpreter (CLI) commands to automate- frequently performed StorageWorks operations.o  - 	The question then is, "Are you experienced?"n   	Apologies to Hendrix.   				Robt   ------------------------------    Date: 26 Jan 2001 11:06:13 -0500* From: young_r@eisner.decus.org (Rob Young)0 Subject: Re: Writes flushed to Disk on Dismount?+ Message-ID: <KJ1Z2q5jeVsW@eisner.decus.org>w  O In article <3A719A97.6090803@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes:dI > My best guess is that if you had an application doing a ton of writes, l > the controllers could cachefH > a lot of blocks without having to start writing them to disk after 45  > seconds.  By setting theH > timeout to 64k second, you are telling the controller "Write these to ! > the disk when you get a chance..% > There's no hurry.  Take your time."  >   5 	But this isn't accurate and which raised my initial '/ 	thread.  What it really means is:  "hold these 4 	writes for 65K seconds before flushing"  of course,8 	the reality is most of those get pushed out in a modest7 	write environment.  But during low I/O times one couldr; 	imagine a write is in cache for a few hours before gettinga 	committed.n   				Rob    ------------------------------  % Date: Fri, 26 Jan 2001 10:21:18 -0600e1 From: "Dave Gudewicz" <david.gudewicz@abbott.com> 0 Subject: Re: Writes flushed to Disk on Dismount?8 Message-ID: <94s7ua$dq9$1@fizban.fizban.pprd.abbott.com>  L "Jan Vorbrueggen" <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de> wrote inJ message news:y4y9vyn1pz.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de...5 > "Dave Gudewicz" <david.gudewicz@abbott.com> writes:  >sH > > When you dismount the disk on an HSJ controller, cache is flushed to disk. I > > On an HSZ or HSG controller, the data remains in cache. The reason is  thatH > > the HSJ uses the MSCP protocol, and a dismount translates to an MSCPJ > > AVAILABLE command. This command results in a cache flush. On a G or Z, thel3 > > protocol is SCSI, and there is no such command.b >hG > I can understand that SCSI doesn't have a straight equivalent to MSCPrF > AVAILABLE. But it seems reasonable that there should be some commandH > that tells the drive to flush its cache, or possibly one that disablesE > it and as a side effect flushes any still remaining cache contents.aI > The HSZ/G port driver (whatever it is called today) should then be ablejI > to translate an AVAILABLE to a series of SCSI commands containing that.7 >8 > Jans  E I agree there "should be" an equivalent AVAILABLE SCSI command.  But,eH according to Ken Bates, there ain't.  I guess there is some truth in theH first "S" in SCSI being "small".  Somehow this "small" made its way into "not so small" systems.n   Dave...    ------------------------------  % Date: Fri, 26 Jan 2001 10:24:17 -0600r1 From: "Dave Gudewicz" <david.gudewicz@abbott.com>i0 Subject: Re: Writes flushed to Disk on Dismount?8 Message-ID: <94s842$dqa$1@fizban.fizban.pprd.abbott.com>  2 "Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515". <Fairfield@SLAC.Stanford.EDU> wrote in message- news:b7tsUmi96Qz$@mccdev.slac.stanford.edu... I > In article <3A70D6F2.3000500@wi.rr.com>, Scott Vieth <svieth@wi.rr.com>s writes:l	 > > Dave:  > > L > > I've been following this little thread and I'm wondering why there is soG > > much worry regarding whether the blocks in the cache are written to  disk. J > > Just because you've dismounted the drive, you can't just go rip it out of > > your ESA or MA cabinet.e >DJ >         You can't?  Why not?  I can do  that with the drives in my SW300J >     &  RA450.   I mean, they _are_ hot-swappable.  Of course, these  are$ >     HSJ-hosted.  not HSZ or HSG... >-J >         But isn't the real issue  whether  you  get all writes completedJ >     and  flushed  to  disk on system shutdown (disk  dismount)  so  thatJ >     you're assured you can power things down and have the  data  intact?J >     I  dunno,  having  to  worry whether the SCSI disk I just dismountedJ >     really had all its  data  written  before  swapping a controller, or+ >     something, makes me pretty nervous...  >JJ > > I've been out-of-the-loop for a while.  Who's KB?  Is it a secret?  If so,uH > > email me offline.  You can trust me to keep your secret.  Ha ha ha!! > @ >         Wouldn't that be Ken Burns, storage guru extrodinaire? >  >             -Ken > --/ >  Kenneth H. Fairfield            |  Internet:i Fairfield@SLC.Slac.Stanford.Edu < >  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924< >  Menlo Park, CA  94025           |  FAX:      650-926-3515L >  ------------------------------------------------------------------------- ----D >  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...  J Ken Burns is a wonderfully talented producer of stories, such as baseball,G jazz (his new project) and the civil war.  If you get  a chance, take ao< look.  PBS not "survivor" stuff.  There is a difference IMO.  B Ken Bates is a wonderfully talented engineer in the storage arena.  , May they both remain in their chosen fields.   Dave...i   ------------------------------  # Date: Fri, 26 Jan 2001 16:50:27 GMTe$ From: Scott Vieth <svieth@wi.rr.com>0 Subject: Re: Writes flushed to Disk on Dismount?( Message-ID: <3A71AB26.2080205@wi.rr.com>  F I visited that web page.  Please tell me we can download that utility 
 somewhere.- I can't believe that we'd have to pay for it.tH Kinda like buying a Viper and then finding out you have to pay extra for floor mats.t  
 -scott :^)   Rob Young wrote:  : > http://www5.compaq.com/products/sanworks/sanmgt/scs.html >  >  > 5 >                          SANworksT Command Scripter. >  > P > The Compaq SANworks Command Scripter application software provides experiencedQ > IT managers with command-level control of Compaq StorageWorksT systems equipped>Q > with HSG60, HSG80, HSZ70 and HSZ80 Array Controllers. Command Scripter works inTQ > a heterogeneous host environment that includes Compaq Tru64 UNIX, Windows 2000,i' > Windows NT, Sun Solaris, and OpenVMS.  > P > With Command Scripter, IT managers can create, edit, and run script files thatJ > contain StorageWorks Command Line Interpreter (CLI) commands to automate/ > frequently performed StorageWorks operations.b > / > 	The question then is, "Are you experienced?"t >  > 	Apologies to Hendrix. > 	 > 				Roby >    ------------------------------   Date: 26 Jan 2001 10:32:59 PSTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)0 Subject: Re: Writes flushed to Disk on Dismount?3 Message-ID: <$33ks5O3TtxX@mccdev.slac.stanford.edu>l  9 In article <009F6AA1.BBBEE21A@SSRL04.SLAC.STANFORD.EDU>, nK     winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr") u     writes: 6 > In article <b7tsUmi96Qz$@mccdev.slac.stanford.edu>, 0     Fairfield@SLAC.Stanford.EDU (Ken Fairfield; .     SLAC: 650-926-2924; FAX: 926-3515) writes:J >>In article <3A70D6F2.3000500@wi.rr.com>, Scott Vieth <svieth@wi.rr.com>      writes:rN >>> I've been out-of-the-loop for a while.  Who's KB?  Is it a secret?  If so,H >>> email me offline.  You can trust me to keep your secret.  Ha ha ha!! >>@ >>        Wouldn't that be Ken Burns, storage guru extrodinaire? > , > Ken Bates, storage guru extraordinare. ...  H         Gads!!!  Of course...  The old  brain  cells are fading fast.  IH     was  actually  "merging" Ken Bates and Burns Fisher.  Burns  is,  ofH     course, a guru in his own right.  At least I didn't say it was BatesH     Fisher or Burns Bates, or even Ken Fisher! ;-p  Apologies to Ken and
     Burns.           -Ken -- 6M  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edu,:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  # Date: Fri, 26 Jan 2001 22:58:33 GMTr( From: Terry Kennedy <terry@gate.tmk.com>0 Subject: Re: Writes flushed to Disk on Dismount?' Message-ID: <G7sL5L.61r@spcuna.spc.edu>   1 Dave Gudewicz <david.gudewicz@abbott.com> writes: G > I agree there "should be" an equivalent AVAILABLE SCSI command.  But, J > according to Ken Bates, there ain't.  I guess there is some truth in theJ > first "S" in SCSI being "small".  Somehow this "small" made its way into > "not so small" systems.d  J   Eh? I have utterly no experience with HS controllers newer than the HSC, but I can comment on SCSI...  J   A SCSI HSx controller almost certainly doesn't just couple a target diskG to the host, even if the config is just a single disk being served to aiJ host - SCSI commands are sent to the HSx, where it may modify them, act onJ them (for example, a read for a block held in cache), or pass them throughH to the disk drive. Further, if in this simple config the HSC didn't lookI at SCSI commands, we wouldn't be having this discussion as it wouldn't bee caching data for that disk.   J   Given that the HSx controller is always listening for SCSI commands, andI the VMS host also always knows when it is talking to a HSx controller, ittJ would be possible to implement a vendor-unique command to force the HSx toI flush its cache for that drive. Or it could interpret the existing START/-J STOP UNIT command as a request to flush the cache. Fortunately, neither ofI these hacks are necessary, as SCSI has included the SYNCHRONIZE CACHE op-nK code (op 35h) and, furthermore, the SCSI spec states "Targets that contain hJ cache memory shall implicitly perform a SYNCHRONIZE CACHE command for the K entire medium prior to executing the STOP UNIT command." so even the START/'L STOP UNIT hack would work. The description of SYNCHRONIZE CACHE reads: "The M SYNCHRONIZE CACHE command (see table 142) ensures that logical blocks in the uK cache memory, within the specified range, have their most recent data valueoJ recorded on the physical medium. If a more recent data value for a logicalG block within the specified range exists in the cache memory than on theyF physical medium, then the logical block from the cache memory shall beK written to the physical medium. Logical  blocks are not necessarily removedcF from the cache memory as a result of the synchronize cache operation."  K   These specs are from 1993 - they're not something new. So, please, beforeIM disparaging SCSI, take a look at what exactly is in there. And Compaq doesn'toK have the excuse of "Vendor X didn't implement this" since they control bothiM the VMS drivers and the code in the HSx's - the disk drives themselves aren'te involved here.  4         Terry Kennedy             http://www.tmk.com5         terry@tmk.com             Jersey City, NJ USAr   ------------------------------   Date: 26 Jan 2001 15:55:34 PSTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515)0 Subject: Re: Writes flushed to Disk on Dismount?3 Message-ID: <bSrqPUuoOdyF@mccdev.slac.stanford.edu>   M In article <3A7199C2.60609@wi.rr.com>, Scott Vieth <svieth@wi.rr.com> writes:w  D > Haven't chatted with you since we were working on stuffing the 4k  > firmware into CIXCDs....  H         Indeed!  And the 4K  stuff  all  worked  out  nicely in the end,     thanks.  :-)   [...]@J > Are you talking about powering down your hosts or the storage cabinet?  J > If it's the storage cabinet, aren't you going to do a "SHUTDOWN OTHER"  K > and "SHUTDOWN THIS" before you power down your storage array?  I bet     kJ > *those* commands would flush your writes to disk.                         H         OK,  sure,  you're   right.    In  controlled  situations,  likeH     anticipating  a scheduled outage, one would definitely use  SHUTDOWNH     on the controllers.  I'd sorta forgotten that.  But to play  devil'sH     advocate:   we   have  our  cluster  and  its  storage  on  a  powerH     conditioner, not a UPS, (for historical  reasons but also because ifH     the power goes away, there's no accelerator to control anyway :-} ).H     We  do suffer unplanned power outages at times.  Now in an unplannedH     power outage, the number of unflushed writes will be proportional toH     the  time  a  write  is  allowed  to  reside  in  cache,  i.e.,  theH     CACHE_FLUSH_TIMER.   In  fact, the algorithm used by the controllersH     probably flushes as soon as the  controller is more-or-less idle, orH     can  squeeze the write in between other activities, so it's unlikelyH     for any write to remain unflushed for 64K  seconds  except  in  veryH     special  circumstances (lots of writes updating the same LBN at veryH     short  intervals  on  a  very  busy  controller).   OTOH,  a shorterH     CACHE_FLUSH_TIMER  guarantees  the writes hit the disk in a  finite,     predetermined time.o  J > Why would you pull a drive out of your storage cabinet (even if they are> > hot-swappable) without deleting the unit and the disk first?  H         Well, sure, you delete the unit and  delete the disk.  But is itH     documented  that either of those commands flush writes to the  disk?H     Perhaps I'm moving the disk to another shelf, or just relabeling it,H     but I want to retain the  data.   Where's  the  guarantee  that  all     writes have been flushed?t  I > Now you're talking about run CSWAP (or whatever the name of the board   I > swap utility is.  I haven't used it in a a while....) Again, I bet that7I > executing the commands to replace a controller board in a redundant     I > pair would flush the caches.                                              !     	Right.  Red herring.  Sorry.n  
     	    -Kenn -- uM  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.EduT:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  % Date: Fri, 26 Jan 2001 20:06:35 +0000 + From: "antonio.carlini" <arcarlini@iee.org>nC Subject: Re: [Q]: Resetting privileged password on a DECserver 200?k' Message-ID: <3A71D8CB.E347A467@iee.org>    Brian Tillman wrote:F > I don't know about other DECservers, but for a DS200, the command isK > DEFINE/SET PRIVILEGED, not SET SYSTEM.  The command to set the privileged   1 I think you're right - I goofed; after many yearst2 of SET PRIV<RETURN>SYSTEM (I hope) I now no longer2 have to manage the beasts. After just eight months& the neural pathways seem to be fading!   Antonio    -- o   ---------------X- Antonio Carlini             arcarlini@iee.orgA   ------------------------------   End of INFO-VAX 2001.053 ************************